Structural Basis for Activation of Human Sirtuin 6 by Fluvastatin

Abstract

Sirtuins are NAD⁺-dependent protein lysine deacylases that are considered attractive drug targets for aging-related diseases. Sirt6 deacetylates, e.g., transcription factors and histone H3, and regulates metabolic processes and stress responses. It has been implicated in lifespan extension and tumor suppression. Sirt6 deacetylase activity can be stimulated with small molecules, and fluvastatin, an FDA-approved synthetic statin, was recently described as a novel Sirt6 activator. We studied the molecular details of this effect on Sirt6 in deacylation assays and by solving a crystal structure of a Sirt6/fluvastatin complex. We find that fluvastatin inhibits Sirt1–3 at higher concentrations but has a unique, activating effect on Sirt6. The complex structure reveals that fluvastatin occupies the Sirt6 substrate acyl channel exit, similar to other, unrelated activator families, providing interaction details that will support the development of potent, druglike Sirt6 activators.

Sirtuins form the conserved histone deacetylases (HDAC) class III of NAD⁺-dependent protein lysine deacylases. They regulate metabolic homeostasis, stress responses, and genomic stability.¹ The seven mammalian isoforms, Sirt1–7, have a conserved catalytic core but differ in localization and substrate preferences.² Sirt6 deacetylates histones and associates with chromatin to regulate gene transcription, telomere integrity, and DNA damage responses.³ SIRT6-knockout mice display lower body size, shorter lifespan, and severe aging-related degenerative phenotypes, such as cancers and metabolic defects.³ Sirt6 expression decreases with age, and SIRT6 overexpression impairs development of several cancer types and extends the lifespan of male mice.^3,4 Therefore, Sirt6 activators are studied as a promising approach for therapy of aging-related diseases.

The sirtuin catalytic domain consists of a large α/β Rossman-fold domain and a small, structure-stabilizing zinc-binding domain.⁵ The two domains are connected by several loops, which form an active site cleft in which both the acylated polypeptide and NAD⁺ cosubstrate are bound. Little variations within the peptide binding cleft and acyl recognition funnel, respectively, cause the isoform-specific substrate sequence and acyl group preference.⁵⁻⁷ During catalysis, NAD⁺ binding causes a conformational change and places its nicotinamide group into the C site, allowing its glycosidic bond to be attacked by the acyl oxygen. The resulting alkylimidate intermediate is then rearranged into a bicyclic intermediate and eventually hydrolyzed to 2′-O-acyl-ADP-ribose and the deacylated polypeptide.^7,8 Among the mammalian sirtuins, Sirt6 has a uniquely wide acyl channel, enabling it to hydrolyze fatty-acetyl groups more efficiently than acetyl group in vitro.⁷⁻⁹ Sirt6-dependent demyristoylation promotes secretion of tumor necrosis factor-α (TNFα) in vivo, but Sirt6 shows also robust deacetylation activity against histones in the context of nucleosomes and chromatin.^3,7,10 Most physiological Sirt6 functions have in fact been linked to its deacetylase activity, especially against histone H3K9ac and H3K56ac.³

Several small molecules have been reported to activate the Sirt6 deacetylase activity.² Certain fatty acids and fatty acid ethanolamides were first found to stimulate Sirt6 at high micromolar concentrations.^11,12 Subsequently, we identified synthetic pyrrolo[1,2-a]quinoxalines as more potent Sirt6 activators,⁹ and more recently the sulfonamide MDL-801 was described as another potent Sirt6 activator.¹³ Complex structures with pyrrolo[1,2-a]quinoxaline-based activators revealed that they bind to Sirt6’s specific substrate acyl binding channel.⁹ The plant flavonoid quercetin and some derivatives, such as cyanidin, occupy the same binding site and also activate the deacetylase activity of Sirt6, albeit with low potency.^14,15 Interestingly, there are also quercetin derivatives that inhibit Sirt6 with promising potency, for example, catechin gallate, and these inhibitors exploit the same Sirt6 site.^14,15 Also trichostatin A (TSA), the most potent known Sirt6 inhibitor (K_i 2–5 μM), inhibits Sirt6 selectively through occupying this site rather than chelating the zinc ion as for HDACI/II inhibition.^16,17 The Sirt6 acyl channel exit thus constitutes an isoform-specific and versatile allosteric regulation site, and information on both, activators and inhibitors, can provide insights helpful for drug development.^9,15

Fluvastatin is a potent HMG-CoA reductase inhibitor FDA approved for treatment of hypercholesterolemia and cardiovascular diseases and with promising effects against hepatocarcinoma and breast cancer cells.^18,19 Fluvastatin was found to activate Sirt6 in “Fluor-de-Lys” (FDL) deacetylation assays potently (EC₅₀ 7.1 μM), while other statins yielded no effect.¹⁹ It can further stimulate Sirt6-dependent deacetylation of H3K9ac and H3K56ac in HepG2 cells.¹⁹ In the present study, we aimed to characterize the Sirt6 binding site and activating effect of fluvastatin to support Sirt6 modulator development. Deacetylation assays confirmed fluvastatin as a scaffold for Sirt6 activator development. A crystal structure of a Sirt6/ADP-ribose/fluvastatin complex revealed as binding site the acyl channel exit, which also accommodates previously identified activators. The compound’s heptenoic acid moiety extends into a distal channel region and indicates a novel binding pocket, and activity assays support the identified binding mode. Our results thus provide molecular insights in Sirt6 activation helpful for further activator development.

Activation of the Sirt6 in vitro deacetylation activity by fluvastatin was reported based on assays with the FdL substrate, whose fluorophore label can cause false positive results.⁵ We therefore examined fluvastatin effects on Sirt6 using FdL substrate as well as an unlabeled peptide substrate representing the physiological substrate site, histone H3 Lys9 (H3K9ac). The FdL assay, performed with a control for artificial fluorophore interaction effects (see Experimental Procedures in the Supporting Information), supported the reported activation (Figure 1a). Mass spectrometry-based assays with an H3K9ac substrate peptide yielded dose-dependent activation, confirming the effect (Figure 1b). The maximum activation was higher than 3-fold, but the compound showed significantly lower potency (EC₅₀ > 250 μM) than previously reported.¹⁹

Figure 1.

Sirt6 activation by fluvastatin and crystal structure of their complex. (a) Effects of fluvastatin on Sirt6 deacetylase activity in the FdL assay (n = 3; error bars: SD). (b) Effects of fluvastatin on the deacylation activity of two Sirt6 constructs in the mass spectrometry-based assay (n = 3; error bars: SD). (c) Overall structure of a Sirt6/ADP-ribose/fluvastatin complex. ADP-ribose is shown as yellow sticks, and fluvastatin is shown in cyan. His133 indicates the active site. (d) Close up of fluvastatin (cyan) in its Sirt6 binding site (interacting residues as gray sticks). 2Fo-Fc electron density for fluvastatin was contoured at 1σ. (e) Scheme of interactions between fluvastatin and Sirt6. (f) Fluvastatin binding site region of a reference structure Sirt6-3-318/ADP-ribose (PDB ID 6XV6; You et al., submitted). The Sirt6 N-terminus (with 2Fo-Fc electron density at 1σ) overlaps with fluvastatin (cyan sticks).

To identify binding site and interaction details of fluvastatin, we first attempted to determine the crystal structure of a Sirt6/ADP-ribose/fluvastatin complex through soaking of crystals of the protein construct Sirt6-3-318 (Experimental Procedures in the Supporting Information) but failed to obtain a complex structure. Since the previously reported activator binding site is located next to the mostly flexible, and in crystal structures rarely observed, Sirt6 N-terminus,^7,9,15 we speculated that this protein region might influence fluvastatin binding and activation. Indeed, activity assays revealed an about 4-fold increase in stimulation of the N-terminally truncated Sirt6(13-355), supporting a potential competition of compound and Sirt6 N-terminus (Figure 1b; potencies cannot be compared since saturation was not reached). Thus, we crystallized the N-terminally truncated construct Sirt6(13-308) in complex with ADP-ribose and soaked the crystals with fluvastatin, yielding a structure of a Sirt6/ADP-ribose/fluvastatin complex. The structure was refined at 2.46 Å resolution to R/R_free values of 15.2/20.5% (Table 1; Supporting Information Table 1). There are two Sirt6/ADP-ribose complexes in the asymmetric unit, and the electron density clearly identified the additional fluvastatin ligand in one of them (Figure 1c,d). The activator is positioned at the exit of the Sirt6-specific acyl channel, with its indole moiety oriented into a rather hydrophobic pocket formed by Phe64/82/86, Ile61, Pro62, and Met136 (Figure 1d,e). The fluorophenyl group and the isopropyl moiety point toward the channel exit and establish contacts to the side chains of Met157, Val70, and Pro80. Fluvastatin’s heptenoic acid moiety interacts with a Sirt6 surface patch comprising Trp71, Glu74, and Lys15. The ligand is bound in the biologically active acid form, rather than the less stable lactone form,²⁰ and extends toward Trp188 to form a hydrogen bond through its carboxyl group (Figure 1d,e).

Table 1. Diffraction Data and Refinement Statistics.

	Sirt6-13-308/ADPr/fluvastatin
data collectiona
space group	P63
cell dim. a = b, c (Å)	91.41, 144.28
resolution (Å)b	48.09–2.46 (2.61–2.46)
unique reflections measured	24 896
I/σIb	10.9 (1.5)
redundancyb	8.7 (8.7)
completeness (%)b	99.9 (99.6)
Rmeasb	0.20 (1.43)
CC1/2b	0.995 (0.536)
refinement
resolution (Å)b	45.70–2.46 (2.52–2.46)
No. reflections	23624
Rwork/Rfree (%)	15.2/20.5
twin fractionsc	0.55/0.45
no. atomsd	4539
B-factorsd
protein	50.6
ligands and solvent	54.3
RMS deviations
bond lengths (Å)	0.01
bond angles (deg)	1.7

^aData collected at BESSY beamline MX 14.1.²¹

^bHighest-resolution shell is shown in parentheses.

^cTwinning was detected through L-tests with POINTLESS (twin law −k, −h, −l), and twin fractions were determined during amplitude-based twin refinement with Refmac.

^dFor details see Supporting Information Table 1.

To analyze how the N-terminal region of Sirt6 can affect the activation by fluvastatin, we compared the complex to a crystal structure of Sirt6-3-318 in the absence of activator (PDB ID 6XV6; You et al., submitted). In this structure, the N-terminus is well-defined by electron density (in some chains within the asymmetric unit) next to the Sirt6-specific acyl channel exit (Figure 1f). Residues 3–10 would overlap with the remote part of the fluvastatin binding site around Trp71 and clash with the compound, rationalizing the more pronounced fluvastatin effect on the N-terminally truncated Sirt6(13-355). It indicates that activators binding to this site either have to be short, to avoid competition with the N-terminus, or add sufficient interactions in this region to replace the N-terminus.

Our structure shows that fluvastatin is accommodated in a wide Sirt6 funnel, which is covered by cofactor binding loop and neighboring helix bundles in other sirtuin isoforms, indicating it should be selective for Sirt6. We indeed find that Sirt6 is uniquely activated by higher concentrations of fluvastatin (Figure 2a). In contrast, the compound caused inhibitory effects on other isoforms, with comparable potency on Sirt1 and 2 and slightly less potent against Sirt3. Since the Sirt6 site for activation is unique, consistent with the unique activation of this isoform, we speculate that the compound can also occupy the quercetin site identified in other isoforms, which is absent in Sirt6.¹⁵ Further studies on other isoforms will be needed, however, to reveal their fluvastatin binding site for reducing these unwanted effects.

Figure 2.

Comparison of fluvastatin on other sirtuins and to other Sirt6 activators. (a) Effects of fluvastatin on deacylation activities of human sirtuin isoforms. (b) Overlay of Sirt6/ADP-ribose complexes with UBCS039 (gray; PDB ID 5MF6) and fluvastatin (cyan). (c) Overlay of Sirt6/ADP-ribose/fluvastatin (cyan) with Sirt3/acetyl-ACS peptide/carba-NAD⁺ (PDB ID 4FVT; carba-NAD⁺ as green sticks). (d) Overlay of Sirt6/ADP-ribose/TSA (gold; PDB ID 6HOY) and Sirt6/ADP-ribose/fluvastatin (cyan). Dashed red lines: H-bonds.

The Sirt6 fluvastatin binding region is exploited by several modulators. Fluvastatin orients its indole group similar to the pyridine portion of the Sirt6 activator UBCS039 but exhibits lower potency. Comparison of their complexes shows that, due to its bulky substituents, fluvastatin cannot enter as deep as UBCS039 into the Sirt6 pocket next to Pro62 (Figure 2b). It thus cannot establish the polar contact to the backbone oxygen of Pro62 seen with UBCS039 and other ligands, which is considered a key interaction.^9,15 Therefore, adding an amino, hydroxyl, or halogen group to the fluvastatin indole might establish a polar contact to Pro62 and enhance Sirt6 binding. The bulky indole substituents in fluvastatin, i.e., the fluorophenyl and isopropyl group, mediate only weaker and nonspecific interactions to Sirt6. Since they seem to prevent deeper binding of the indole group, reducing their size might improve the potency against Sirt6; and it would simultaneously decrease the nanomolar affinity to the main fluvastatin target, HMG-CoA reductase, where these groups contribute significant hydrophobic interactions.²² Alternatively, they could be modified to insert polar interactions for compound improvement or the contact to Trp71 seen with the diaminomethylphenyl group of TSA (see below).¹⁷ The other way around, UBCS039 binds to Sirt6 coplanar with fluvastatin but lacks the hydrogen bond to the remote site, to Trp188. Adding an extended substituent to the pyrrolo[1,2-a]quinoxaline group of UBCS039 to exploit this binding site region could improve binding of the UBCS039 compound family further.

The Sirt6 modulators exploiting the Sirt6 acyl channel exit overlap with the binding site for the nicotinamide portion of NAD⁺ (“C site”). They thus appear to bind after dissociation of the nicotinamide moiety in the first catalytic step, as established for the Sirt1 inhibitor EX527,²³ and to modulate the stability of the Sirt6/product complex. Although fluvastatin enters less deep into the pocket lined by Pro62, it would still overlap with the C site (Figure 2c). Thus, fluvastatin appears to activate Sirt6 through the same mechanism. Comparing the Sirt6/ADPr/fluvastatin complex to Sirt6/ADPr (PDB ID 6XV6) illustrates the clash with the protein’s N-terminus (see above; Figure 1d) and leads to a flip of the Trp188 side chain for its polar interaction with the activator. Other residues around the compound show no or only slight rearrangements. There are no major conformational differences in other protein regions and it remains to be established how the compound elicits activation.

Since there are also ligands for the fluvastatin-accommodating site that inhibit rather than activate Sirt6, the described binding details can also be exploited for inhibitor development. The potent, isoform-selective Sirt6 inhibitor TSA occupies Sirt6’s acyl channel with its aliphatic linker and diaminomethylphenyl group.^16,17 The TSA hydroxamate is positioned deep in the Sirt6 C-site and mimics, and improves beyond, the interactions between nicotinamide and Sirt6.¹⁷ An overlay of Sirt6/ADP-ribose/fluvastatin and Sirt6/ADP-ribose/TSA matches the fluvastatin indole group with the aliphatic TSA linker (Figure 2d). Attaching a hydroxamate to fluvastatin should improve its affinity, and possibly switch its effect to inhibition, by exploiting the C site with this deeply entering moiety. Importantly, such a chimera would lose the shape and conformation that allows TSA to inhibit Class I/II HDACs with ∼100-fold higher potency than Sirt6 and thus could provide an avenue to the development of highly specific Sirt6 inhibitors.

Despite the physiological, and potential therapeutic, relevance of Sirt6, only few Sirt6 modulators have been discovered, and they tend to suffer from limited efficacy, specificity, or other drug properties.^1,2,24 Fluvastatin was identified as a potent Sirt6 activator through assays with the potentially artifact-causing, fluorophore-labeled FdL substrate (EC₅₀ 7.1 μM).¹⁹ We could confirm the Sirt6 activating effect of fluvastatin using a H3K9-derived, unlabeled peptide substrate, albeit with lower potency. The previously reported cellular effect of low fluvastatin doses on histone H3 acetylation might actually be indirect, through increased nuclear translocation of Sirt6.¹⁹ Nevertheless, fluvastatin and its binding interactions identified here provide additional information for the development of improved Sirt6 modulators. Its hydrogen bond with a hydrophilic patch on the target surface should allow improvement of other ligands with respect to affinity as well as solubility. This includes MDL-801, a more recently identified, micromolar Sirt6 activator (EC₅₀ 10 μM), which was reported to bind to a neighboring, more exposed side¹³ but was now shown to share significant interactions with UBCS039 and fluvastatin.²⁵

Statins can induce the sterol regulatory element-binding protein, a key regulator of cholesterol homeostasis, and such an effect has also been attributed to the Sirt6 histone deacetylation activity.^26,27 Thus, Sirt6 modulation might contribute to physiological fluvastatin effects and the molecular details of the Sirt6/fluvastatin interaction thus could also help to improve fluvastatin specificity for its intended target, HMG-CoA reductase.

In summary, our results confirm fluvastatin as a Sirt6 activator and reveal its binding mode. Our results provide a novel scaffold for exploiting the allosteric Sirt6 site as well as helpful data for the design of improved Sirt6 activators and inhibitors.

Glossary

Abbreviations

FdL

Fluor-de-Lys

Sirt

sirtuin

H3K9ac

acetylated histone H3 Lys9

HDAC

histone deacetylase

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.0c00407.

• Experimental procedures; refinement statistics details (PDF)

Author Contributions

W.Y. and C.S. designed the study, analyzed the data, and wrote the manuscript. W.Y. performed the experimental work. C.S. supervised the project and acquired funding.

The authors declare the following competing financial interest(s): C.S. is a SAB member of Ovibio.

Notes

Coordinates and diffraction data for the Sirt6/fluvastatin complex have been deposited with the wwPDB (www.pdb.org) under accession code 6ZU4.