Bioluminescence assay of lysine deacylase sirtuin activity

SUMMARY

Lysine acylation can direct protein function, localization, and interactions. Sirtuins deacylate lysine toward maintaining cellular homeostasis, and their aberrant expression contributes to the pathogenesis of multiple conditions, including cancer. Measuring sirtuins’ activity is essential to exploring their potential as therapeutic targets, but accurate quantification is challenging. We developed “SIRTify”, a high-sensitivity assay for measuring sirtuin activity in vitro and in vivo. SIRTify is based on a split-version of the NanoLuc luciferase consisting of a truncated, catalytically inactive N-terminal moiety (LgBiT) that complements with a high-affinity C-terminal peptide (p86) to form active luciferase. Acylation of two lysines within p86 disrupts binding to LgBiT and abates luminescence. Deacylation by sirtuins reestablishes p86 and restores binding, generating a luminescence signal proportional to sirtuin activity. Measurements accurately reflect reported sirtuin specificity for lysine-acylations and confirm the effects of sirtuin modulators. SIRTify quantifies lysine deacylation dynamics and may be adaptable to monitoring additional post-translational modifications.

In brief

Van Scoyk et al. report a high sensitivity assay for measuring sirtuin activity. Using modified NanoBit, where lysine acylation disrupts luminescence, deacylation by specific sirtuins restores luminescence, allowing accurate measurements of activity. The platform is used to explore therapeutics for sirtuin modulation and can be adapted to monitor other modifications.

Graphical abstract

INTRODUCTION

Reversible acylation of protein lysines regulates cellular processes, including transcription, metabolism, and signaling, to influence critical cell fate decisions, such as differentiation and apoptosis. To date, at least twenty different lysine acyl modifications have been reported, including acetylation, crotonylation, succinylation, and glutarylation.¹ Although recent data indicate that, with the exception of acetylation, non-enzymatic reactions account for the bulk of lysine acylations,^1–6 deacylation depends mainly on the activity of a single family of enzymes, termed sirtuins.^2,3

The sirtuin family consists of seven members (SIRT1–7) with diverse and partially overlapping substrate specificity, cellular location, and function. Sirtuins are evolutionarily conserved NAD⁺-dependent lysine deacylases implicated in several fundamental biological processes.^7,8 Dysregulation of sirtuins is associated with a range of pathological conditions, including diabetes, inflammatory disorders, neurodegenerative diseases, and cancer. Considering their impact on key cellular processes, it is imperative to develop effective tools to characterize and monitor acyl modifications by sirtuins in vitro and in vivo to delineate their potential as therapy targets.^7,9–14

Several assays have been developed to measure sirtuin activity, but only a few are commercially available.^15,16 The most widely used test is FLUOR DE LYS (FDL), which is a two-step assay that probes deacylation of an acyl-lysine peptide conjugated to the fluorophore aminomethylcoumarin (AMC). The peptide-fluorophore is “essentially nonfluorescent” at the excitation (390 nm) and emission (460 nm) compared to free AMC.¹⁶ Following deacylation, trypsin is added as a developing agent to cleave the AMC from the deacylated peptide. The resulting fluorescence signal is proportional to sirtuin activity. The utility of the FDL assay is limited by its tendency to produce false positive results from nonspecific interactions of the fluorophore with the sirtuin-modulating compounds^17–22 and incompatibility with more complex systems, including cell lysates, intact cells, and living organisms. Various other sirtuin assays have been reported that use chemical fluorescent and molecular self-assembly probes, radioisotope-labeled histones, nicotinamide release, HPLC, or mass spectrometry.^23–26 These assays are neither widely used nor readily available, reflecting a range of limitations related to chemical stability, selectivity, expense, and/or toxicity.^{17–21,27–29} For example, a previous report by Kawaguchi et al., describes a fluorescent probe used to test the activity of SIRT1 in living cells showing a correlation between sirtuin inhibition and fluorescence activity; however, the probe was limited to SIRT1 activity and sirtuin activators were not tested in live cells.²⁵

To address the limitations of current sirtuin assays, we developed SIRTify, a luminescent assay for measuring and imaging lysine-deacylase activity in intact cells and in vivo. SIRTify is based on a split version of NanoLuc, an engineered luciferase from the deep-sea shrimp Oplophorus gracilirostris.³⁰ In the split-NanoLuc complementation system, removal of the C-terminal β strand of the β-barrel domain renders NanoLuc catalytically inactive.³⁰ Acylation of the two lysines present within the small β strand peptide fragment of split-NanoLuc reduces the complementation-based activity of the split-NanoLuc to very low levels, while deacylation by sirtuins restores luciferase activity. We demonstrate that SIRTify accurately measures sirtuin activity in cell-free systems, in vitro and in vivo, and is useful for identifying sirtuin activity modulators without many of the issues seen in these complex systems with other assay formats.

RESULTS

Biochemical characterization of acylated peptides

NanoLuc is an engineered luciferase consisting of 11 antiparallel strands forming a β-barrel that is capped with four α helices. The first split-NanoLuc complementation system demonstrated that removal of the C-terminal β strand of the 10-stranded β-barrel renders NanoLuc catalytically inactive.³⁰ Sequence optimization of the C-terminal peptide identified a series of 11 amino acids peptides spanning five orders of magnitude in affinity for the large, truncated fragment of NanoLuc (LgBiT).³⁰ From this series, we selected peptide “86” (p86) because of its high affinity to LgBiT (K_D = 0.7 × 10⁻⁹ M) and because it contains two lysines located at position eight and nine of the peptide (NH₂-VSGWRLFKKIS-OH).³⁰ We hypothesized that ε-acylation of these lysines would reduce affinity to LgBiT, preventing reconstitution of active NanoLuc (Figures 1A and 1B). Further, removal of the respective acylation by a sirtuin should restore native p86, allowing for the reconstitution of active luciferase and corresponding luminescence.

Figure 1. Design strategy of the luminescence-based lysine deacylase assay.

(A) Schematic representation of the split-NanoLuc modified system adapted for detection of sirtuin activity by acylation of the lysine residues with peptide p86. NAD⁺ - nicotinamide adenine dinucleotide; NAM – nicotinamide; OAADPr – O-Acetyl-ADP-Ribose.

(B) Structural representation of split-NanoLuc luciferase system, consisting of LgBiT (β-barrels 1–9) and p86 11-aa peptide (β-barrel 10). Sequences of 13-aa native peptide (NP), and p86 peptides.

(C) Titration of LgBiT with p86 and lysine-acylated p86 peptides. K_D values were estimated by fitting data to one site-specific binding model on GraphPad.

(D) Kinetic parameters of furimazine with LgBiT and acylated peptides were estimated by fitting data to a hyperbolic equation.

(E) Table of kinetic parameters shown in (C) and (D). Results in (C)–(E) are from independent experiments performed three times. Data show means with standard deviation.

As a proof of concept, we compared activity for unmodified, single, or dually acetylated and succinylated p86. Single acetylation or succinylation at either K8 or K9 resulted in only a minor decrease of 1.1 and 1.7-fold in luminescent signal, respectively (Figure S1). In contrast, dual acetylation or succinylation at K8 and K9 resulted in a 5.8 and 7.5-fold decrease in signal. As the dual lysine modification was more effective in decreasing luminescence, we modified p86 peptides at both K8 and K9 for a series of acyl modifications—acetyl, crotonyl, succinyl, and glutaryl—to produce p86-Acetyl_8,9, p86-Crotonyl_8,9, p86-Succinyl_8,9, and p86-Glutaryl_8,9, respectively, which we collectively termed p86-Acyl_8,9 peptides.

To determine changes in binding affinity induced by acylation, we measured the dissociation constants (K_D) of LgBiT and p86-Acyl_8,9 interactions by titration. We first determined the K_D of p86 by fitting titration data to a standard model and found it comparable to published data (3.80 × 10⁻⁹ M vs. 0.7 × 10⁻⁹ M reported by Dixon et al.³⁰). We next determined K_D values for p86-Acyl_8,9 peptides (Figures 1C and 1E). Lysine acylation consistently decreased binding affinity for the large split-NanoLuc fragment of all p86-Acyl_8,9 peptides. Compared to p86, p86-Acyl_8,9peptides showed between ~175-fold (p86-Acetyl_8,9) and ~1,000-fold (p86-Glutaryl_8,9) reduced affinity to LgBiT. Next, we measured kinetic parameters, including turnover number (kcat) and Michaelis constant (K_M) of the p86-Acyl_8,9 peptides relative to native p86. Under our assay conditions, K_M of p86 with LgBiT was more than 60-fold lower than that of p86-Acyl_8,9peptides (0.03 μM vs. 1.25–1.88 μM, respectively). Accordingly, the catalytic efficiency of LgBiT expressed as (kcat/K_M) decreased from 3.12×10⁴ μM⁻¹s⁻¹ for p86 to 2.10–4.08×10² μM⁻¹s⁻¹ for p86-Acyl_8,9 peptides (Figures 1D and 1E). Our results demonstrate that, although acylated peptides can still bind to LgBiT and enable substrate conversion, this activity is significantly reduced compared to native p86 peptide.

We tested whether sirtuin activity and specificity toward the p86-Acyl_8,9 peptide substrates were maintained and whether luciferase activity was restored in the presence of sirtuins. Of the SIRT1–7 enzymes, we chose to focus on SIRT1, 2, 3, and 5 due to the limited activity of SIRT4, 6, and 7 (Figure S2). In the absence of sirtuin enzymes, all p86-Acyl_8,9 peptides showed less activity than native p86 (Figure 2A). We observed that SIRT1, 2, 3, and 5 restored the activity of modified p86-Acyl_8,9 peptides in a manner consistent with their reported substrate specificities (Figure 2B). Specifically, incubation of p86-Acetyl_8,9 with recombinant SIRT1, 2, or 3 restored signal intensity close to that of unmodified p86, confirming deacetylation activity. In contrast, incubation of p86-Acetyl_8,9 with SIRT5 showed only low luminescent activity, suggesting that SIRT5 is unable to remove the acetyl modification. This is consistent with several recent reports that found SIRT5 to be a weak deacetylase,^5,6,10 although this is in contrast with an early report.³¹ For p86-Crotonyl_8,9, SIRT1 and 2 also restored signal but not as efficiently as for p86-Acetyl_8,9. SIRT3 and SIRT5 did not restore the signal for p86-Crotonyl_8,9; for SIRT3, there is one report of decrotonylation activity which requires additional validation.³² Incubation of p86-Succinyl_8,9 and p86-Glutaryl_8,9, with SIRT5 resulted in the restoration of signal, while SIRT1–3 did not, again in agreement with reported substrate specificities.⁸

Figure 2. Performance of SIRTify assay in a cell-free system.

(A) Complementation of unmodified p86 with LgBiT results in luminescence which is significantly decreased for p86-Acyl peptides (-Acetyl, -Crotonyl, -Glutaryl, -Succinyl) modified at K8 and K9.

(B) The SIRTify assay reveals each sirtuin specificity for different acylations.

(C) Michaelis-Menten of kinetic parameters of relative kcat and K_M for furimazine with sirtuins.

(D) Table of kinetic parameters determined from curve fits confirms selectivity. Results in (A–D) are from three independent experiments. Data show means with standard deviation p values were calculated using one-way ANOVA. *p < 0.033, **p < 0.002, ***p < 0.001.

Sirtuin deacylation kinetics of p86-acylated peptides

To measure the kinetics of SIRT1, 2, 3, and 5 deacylation, we incubated the sirtuins with acylated p86 peptides (Figures 2C and 2D). Steady-state deacylation rates were determined by measuring luminescence every 5 s for 10–15 min. Comparing deacylation types, SIRT1 and 2 showed k_cat values that were, respectively, ~24 and ~15-fold higher for deacetylation compared to decrotonylation. The catalytic efficiencies (k_cat/K_M) of SIRT1 and 2 were ~4– to 6-fold greater for deacetylation than for decrotonylation. For SIRT5, the k_cat values were 5-fold higher for desuccinylation compared to deglutarylation, while, in contrast, the k_cat/K_M were ~8-fold lower for desuccinylation compared to deglutarylation due to differences in K_M. Comparing between enzymes, for deacetylation, the k_cat value for SIRT3 was similar to SIRT2, while SIRT1 had a k_cat value ~4– to 11-fold higher than SIRT 2 and 3. SIRT1 had ~5 and 11-fold higher k_cat/K_M compared to SIRT2 and 3, respectively. For decrotonylation, SIRT1 had a k_cat value ~2-fold higher and a k_cat/K_M value ~4-fold higher than SIRT2. Together, these results indicate that sirtuins display specificity and differential catalytic efficiency that corresponds to deacylation of the modified p86 peptide.

Using SIRTify to screen for sirtuin inhibitors

Given the involvement of sirtuins in diverse disease states, including cancer, neurodegeneration, cardiovascular disease, and diabetes, there is considerable interest in developing sirtuin inhibitors and sirtuin activating compounds (STACs).¹² To verify that SIRTify is suitable for the identification of such compounds, we tested a set of well-characterized inhibitors, including EX527 (SIRT1 inhibitor), SirReal2 (SIRT2 inhibitor), 3-TYP (SIRT3 inhibitor), and SIRT5 inhibitor 1 (S5I1). For initial experiments, we selected inhibitor concentrations higher than all reported IC₅₀ values (100 μM for 3-TYP and 1 μM for all other inhibitors) (Figure 3A). Incubation with inhibitor reduced the deacylation-associated signal increase for p86-acetyl_8,9 in the case of SIRT1–3, for p86-Crotonyl_8,9 in the case of SIRT1–2, and for p86-Succinyl_8,9 and p86-Glutaryl_8,9 in the case of SIRT5, demonstrating enzyme inhibition for all deacylase activities studied, confirming published inhibitor data. However, we found the IC₅₀ for SIRT1–3 inhibitors to be higher than previously reported (Figures 3C; Table 1).^33–35 We postulated that the higher IC₅₀ of EX527, SirReal2, and 3-TYP may reflect the higher concentration of NAD⁺ used in our assay (250 μM compared to 170 μM)³³ and tested additional NAD⁺ concentrations^33–35. Indeed, using a lower concentration of NAD⁺ decreased the IC₅₀ for SIRT1; however, it remained slightly higher than previously published data (Figure S3). Comparing IC₅₀ values between deacylase types for SIRT5 inhibition, interestingly, we measured an approximately 3-fold lower IC₅₀ (0.047 μM) for inhibition of SIRT5’s deglutarylase activity compared to its desuccinylase activity (0.15 μM). S5I is the most potent SIRT5 inhibitor reported thus far, but published data are based exclusively on desuccinylation. The current standard, FDL assay, does not measure inhibition of SIRT5’s deglutarylase activity; however, comparison between SIRT5’s desuccinylase activity determined from SIRTify and from FDL reported values are within error (Table 1, Figure S4).³⁶

Figure 3. Use of SIRTify to screen sirtuin inhibitors in a cell-free system.

(A) Specific inhibition of sirtuin activity using small molecule inhibitors. SIRT1; Selisistat (EX527) 1 μM, SIRT2; SirReal2 1uM, SIRT3; 3-TYP 100 μM, SIRT5; SIRT5 Inhibitor 1 (1 μM).

(B) SIRT5 H158Y mutant activity vs. wild-type SIRT5.

(C) IC₅₀ of specified sirtuin inhibitors. IC₅₀ values were estimated by fitting data to nonlinear regression using log(inhibitor) vs. normalized response-variable slope and specified in Table 1. Results in (A)–(D) are from three independent experiments. Data show means with standard deviation. p values were calculated using two-way ANOVA with Tukey’s method of adjustment for multiple comparisons. *p < 0.033, **p < 0.002, ***p < 0.001.

Table 1.

Comparison of IC₅₀ values identified from SIRTify to previously published findings

Sirtuin	Inhibitor	Acylation	IC50 (μM)	IC50 (μM) Pub
SIRT1	Selistat (EX527)	Acetyl	1.46 ±0.093	0.038
		Crotonyl	0.365 ± 0.084	–
SIRT2	SirReal2	Acetyl	0.276 ± 0.031	0.14
		Crotonyl	0.207 ± 0.051	–
SIRT3	3-TYP	Acetyl	24.4 ± 5.89	16
SIRT5	SIRT5	Succinyl	0.102 ± 0.008	0.11
	Inhibitor I (S5I1)	Glutaryl	0.047 ± 0.002	–

References for Selisistat,²⁹ SirReal2,³⁰ 3-TYP,³¹ SIRT5 Inhibitor 1.³³

Histidine 158 (H158) is critical for sirtuin catalytic activity.^31,37,38 Replacement of H158 with tyrosine results in a catalytically inactive mutant.^31,37,38 We measured SIRT5-H158Y activity toward p86-succinyl_8,9 and p86-Glutaryl_8,9. In contrast to wild-type SIRT5, SIRT5-H158Y did not produce any luminescent signal, validating specificity (Figure 3B). In aggregate, these data demonstrate that SIRTify accurately measures inhibitor effects on sirtuin activity and has utility for identifying sirtuin-modulating compounds.

Measuring sirtuin activity in cells

Currently, available assays for measuring sirtuin activity in cells have significant limitations, including toxicity, availability, complexity of handling due to the need for special safety precautions, and inactivity in many experimental conditions.^23–25,39 To test whether the SIRTify assay can be adapted for use in cells, we stably expressed LgBiT in HepG2 hepatocarcinoma cells and KG1α acute myeloid leukemia cells and focused on SIRT5 (Figures 4A and 4B). As SIRT5 is primarily located in the mitochondria, we added a mitochondrial targeting sequence (MTS) to LgBiT (LgBiT-MTS) (Figure 4B). As SIRT5 is the main, and possibly the only, mammalian desuccinylase and deglutarylase, we predicted that p86-Succinyl_8,9 and p86-Glutaryl_8,9 deacylation activity would be proportional to SIRT5 expression and/or activity. We found that luminescence correlated with p86 concentration. As previously observed, p86-Succinyl_8,9 had a significantly lower signal as previously observed; however, this change was less dramatic than within the cell-free systems. We first tested whether we can measure changes in SIRT5 activity by NRD167, a cell-permeable S5I1 derivative, or UBSC039, a prospective SIRT5 activator.^36,40 LgBiT-expressing cells were treated with NRD167 or UBSC039 for 2 or 24 h and then incubated with p86, p86-Succinyl_8,9, or p86-Glutaryl_8,9 in lysis buffer for 2 h while gently rocking (Figure 4C). Luminescence was measured following the addition of the luminescent substrate furimazine. Treatment with NRD167 decreased the luminescent signal with both succinyl and glutaryl-modified peptides in the KG1α cell line, whereas UBSC039 produced a signal increase in only the HepG2 line. Next, we tested whether modifying SIRT5 expression would alter luminescence. We overexpressed (OE) SIRT5 in our LgBiT stably expressing cells and measured luminescence as described previously (Figure 4D). SIRT5 OE increased luminescence for both p86-Succinyl_8,9 and p86-Glutaryl_8,9 peptides, confirming that SIRTify is sensitive to changes in SIRT5 expression.

Figure 4. Imaging and quantification of SIRT5 activity in lysed cells.

(A) Schematic of the pCDH vector transcriptional cassette.

(B) Fluorescence imaging of HepG2 cells with stable mitochondrial LgBiT expression (green), nuclei are labeled with Hoechst (blue), and mitochondria with Mitotracker (red).

(C) HepG2 cells (left) and KG1α cells (right) were treated with the SIRT5 prodrug NRD167 (50 μM) or UBSC0O39 (100 μM) for 2 h or 24 h, respectively, before cells were incubated with peptide and lysed.

(D) Left, analysis of SIRT5 expression in Hep2G cells and KG1α cells. Right, immunoblot quantification.

(E) HepG2 (left) and KG1α (right) with SIRT5 OE were lysed and treated with peptide for 2 h. Results in (A)–(D) are from six independent experiments. Data show means with standard deviation. p values were calculated using two-way ANOVA with Tukey’s method of adjustment for multiple comparisons. *p < 0.033, **p < 0.002, ***p < 0.001.

We next tested whether SIRTify can be adapted for live-cell measurement of SIRT5 activity. As the utility of a cellular assay is dependent on sufficient cell permeability and distribution, we initially optimized cellular uptake by modifying p86 through C-terminal addition of four arginine residues (p86-R₄) (Figure S5A).⁴¹ To test whether p86-R₄ and its derivatives are cell-permeable, we incubated the cell lines with furimazine and graded concentrations of p86-R₄, p86-Succinyl_8,9-R₄, or p86-Glutaryl_8,9-R4. Luminescence was directly proportional to p86-R4 concentration, while p86-Succinyl_8,9-R₄ and p86-Glutaryl_8,9-R4 had a much lower signal (Figure S5B). We first tested whether modification of SIRT5 activity by NRD167 or UBSC039 would modify the deacylase activity of SIRT5 for p86-Succinyl_8,9-R₄ and p86-Glutaryl_8,9-R4 in intact cells. We observed that, consistent with our results on lysed cells, inhibition of SIRT5 resulted in decreased luminescent signal in both cell lines, while UBSCO39 showed an increase only in the HepG2 cell line (Figure 5A). In SIRT5 OE cell lines, we observed increased luminescence in HepG2 and KG1α cells, demonstrating that the utility of the SIRTify assay extends to cell-based screening (Figure 5B).

Figure 5. SIRT5 activity in live cells.

(A) HepG2 cells (left) and KG1α cells (right) were treated with the SIRT5 inhibitor NRD167 (50μM) or UBSCO39 (100μM) for 2 h or 24 h, respectively, before incubation with peptide for 2 h.

(B) HepG2 (left) and KG1α (right) with SIRT5 OE were treated with peptide for 2 h. Results in (A) and (B) are from six independent experiments. Data show means with standard deviation. p values were calculated using two-way ANOVA with Tukey’s method of adjustment for multiple comparisons. *p < 0.033, **p < 0.002, ***p < 0.001.

Detecting sirtuin activity in vivo

To determine the utility of SIRTify for measuring sirtuin activity in vivo, we injected HepG2 LgBiT-MTS expressing (right) and parental HepG2 (left) cells into opposing flanks of NRG mice (Figure 6A). After the tumor reached ≥5 mm size, p86-R4 and furimazine were injected intratumorally, and images were acquired. A strong signal was observed only in flanks injected with HepG2-MTS-LgBiT tumors (Figure 6A). Zebrafish models are a convenient approach to high throughput in vivo drug screens. Their major advantage over biochemical and cell line-based screens is that they permit both whole-organism and tissue-specific analysis, potentially accelerating the process of drug development and validation.⁴² To test whether SIRTify may be adapted to a zebrafish-based system, we injected purified LgBiT protein with or without p86-R4 peptide into one-cell stage zebrafish embryos, then incubated them with either furimazine or endurazine, a recently reported alternative substrate for live-cell detection that allows for a steady release of furimazine (Figure S6). A strong signal was observed in embryos co-injected with p86-R4 and LgBiT but not in embryos injected with LgBiT alone. Next, we tested whether we could measure endogenous Sirt5 activity in zebrafish. After injecting LgBiT and p86-Succinyl_8,9-R4, we observed a slight increase in luminescence over the following 90 min, suggesting that endogenous desuccinylation activity exists at a low level in early zebrafish embryos and is detected by SIRTify (Figure 6B). Prior studies looking at larval stage zebrafish investigating gain-of-function or loss-of-functions models of Sirt5 showed that Sirt5 could substantially contribute to protein succinylation at 7 days post fertilization, suggesting that the modest desuccinylation activity we observed in the first few hours after fertilization might be enhanced at later life stages.⁴³

Figure 6. Applicability of SIRTify within in vivo systems.

(A) Schematic of mouse experiment (left), HepG2 parental cells were injected into the left flank of an NRG mouse, while HepG2 LgBiT-MTS cells were injected into the right flank. P86 and furimazine were injected into the tumors, and luminescence was measured. Quantification of luminescence (right).

(B) Schematic of zebrafish experiment (left), purified LgBiT with p86 or p86-Succinyl_8,9-R4 was injected into zebrafish embryos and incubated in the presence of endurazine. Luminescence was measured approximately every 45 s for 90 min.

DISCUSSION

Here, we describe a split-luciferase system that reports sirtuin activity and specificity with high fidelity to benchmarks. The current standard to measure sirtuin activity is the fluorometric system FDL, a two-step process where trypsin is added to cleave the fluorescent AMC conjugate from the deacylated peptide. In cell-free assays and cellular lysates, trypsin may degrade the sirtuin of interest or sirtuin regulatory proteins, a possible explanation for artifacts and lack of specificity.^17,18 We designed SIRTify to overcome this shortcoming, as no second step is required to generate the signal. Additionally, the FDL assay is limited to measuring deacetylation and desuccinylation activity, as there is currently no assay available for other modifications, such as glutaryl and crotonyl. SIRTify not only reproduced previously published results on SIRT1, 2, 3, and 5 specificities for deacetylation and desuccinylation but also detected decrotonylase and deglutarylase activity, allowing for an easy and quantitative comparison of substrate specificity. It is possible that the same approach could be extended to additional sirtuin activities such as demyristoylation or demalonylation.

Similar to FDL, assays based on radiolabeled histones, nicotinamide release, fluorescence polarization, or LC-MS are incompatible with live cell measurements.^27–29 Multiple reports have described activity-based chemical probes to measure sirtuin activity in protein mixtures and cell lysates.^44,45 However, additional work is needed to increase the cell permeability, selectivity, and sensitivity of sirtuin activity-based probes (ABPs) before these can be valuable tools to measure sirtuin activity in vitro. Here, we have demonstrated that SIRTify p86 acyl peptides are cell permeable and can measure changes to SIRT5 activity by either chemical modulators or modifying SIRT5 levels in multiple cell lines. Fluorescent probes have also been described to measure intracellular sirtuin activity.^23,25 One approach relies on the constitutive expression of an EGFP mutant with a non-canonical acetyl-lysine. Where these approaches do not allow for temporally resolved studies (the acetyl-GFP substrate is continuously produced) and are limited to deacetylation activity only, SIRTify can be used for short- or long-term measurements and can detect additional acyl modifications, including succinyl and glutaryl.

The 11-amino acid “p86” does not match any known physiological sirtuin substrate and there is some evidence that sirtuins may preferentially recognize lysines within a specific sequence context; however, it is challenging to identify a native substrate that is amenable to SIRTify. Reports focused on SIRT2 and SIRT3 failed to demonstrate a clear consensus in the amino acid sequences surrounding the acylated lysine residue.^46,47 Likewise, for SIRT5, no specific sequence has been identified, although certain patterns for acylations are beginning to emerge.^48,49 Differences for preferred substrate acyl groups are caused by binding of the acyl moiety to an active site channel within the respective sirtuin. Although sirtuins share a conserved catalytic core of ~275 amino acids, differences within the binding cleft configuration distinguish between sirtuins and are central to substrate specificity.^50,51 The selectivity for acylation is supported by our SIRTify results, as the sirtuins we tested not only recognized the small p86 peptide but also maintained specificity toward the acyl modification. For instance, SIRT5 has very weak deacetylase activity and SIRT3 has activity, while the opposite is known for desuccinylase activity, which we also observe. Nevertheless, it is unlikely that the small p86 peptide imparts substrate specificity per se, and therefore it is likely that other deacetylases are active toward the acylated small peptide. Specificity and activity need to be established experimentally for each deacylase.

Another potential confounding factor is sirtuin subcellular localization. SIRT1 is mainly nuclear, SIRT2 is localized to the cytosol, while SIRT3 and SIRT5 are primarily located within the mitochondria. Additionally, subcellular localization of sirtuins can vary during development, in response to stimuli, and in different cell types.^52,53 For example, SIRT5 is primarily located within the mitochondria but has also been observed within the cytoplasm and nucleus. Investigation has focused mainly on the mitochondrial function of SIRT5 and its role in metabolism. However, recent studies have extended the scope to include the role of SIRT5-mediated histone desuccinylation and its impact on disease.⁵⁴ We have demonstrated that LgBiT can be targeted to the mitochondria with little impact on cellular health. Other modifications, e.g., to promote nuclear localization, may be helpful to elucidate cell compartment-specific SIRT5 functions.

The recognized importance of sirtuins in cellular homeostasis, aging, and disease has increased interest in developing therapeutic modulators of sirtuin activity. However, the development of such compounds faces major challenges, including limited target specificity and potency.⁴⁷ We have shown that SIRTify measurements correlate well with published data,⁵⁵ and we demonstrated that SIRTify is highly selective, allows for the determination of steady-state kinetic measurements, and is readily adaptable for high-throughput screening, which could facilitate the discovery and characterization of therapeutic compounds. Establishing a full dynamic range, further assessing sensitivity, and determining Z-factor values are necessary parameters to adapt SIRTify for high-throughput screening and is a focus of ongoing studies.

Our data indicate the possibility of adapting SIRTify to measure additional post-translational modifications (PTMs), such as ubiquitination and methylation. This may allow quantifying the activity of deubiquitinating enzymes (DUBs) or lysine demethylases (KDM), respectively. DUBS and KDMs are essential regulators of key cellular processes and are involved in autoimmune disorders, cancer, and neurodegeneration.^56,57 In addition to lysines, p86 contains two serines in positions 2 and 11. Common modifications that occur on serine are O-linked glycosylation, methylation, acetylation, and phosphorylation, which can regulate catalytic activity.⁵⁸ Alternatively, additional peptides, characterized by Dixon et al., comprise alternative sequences, such as peptide 78, which has two asparagine residues added on the N and C terminus. Asparagine modifications include phosphorylation, hydroxylation, and N-linked glycosylation.^30,59,60 It may be possible to adapt the SIRTify design to measure the activity of these enzymes in a manner like that shown here for sirtuins.

Limitations of the study

While SIRTify provides many advantages over conventional sirtuin activity assays, limitations remain. It is unknown how the activity of recombinantly expressed and purified sirtuins compares to their activity in situ or how cellular localization of the sirtuins, p86 peptides, or LgBiT may affect results. We have shown that expression of LgBiT tagged with a mitochondrial localization signal is tolerated and may be adapted to other cellular compartments. Peptides with cellular localization signals may also improve the SIRTify signal within cells. Secondly, while we demonstrated the ability of SIRTify to measure deacylation activity within in vivo systems, endogenous SIRT5 activity appeared to be low in the early zebrafish embryos tested. While sirtuins are expressed in zebrafish, there is limited information about their activity, especially SIRT5, during the early developmental stages. It is possible that SIRT5 activity is low during the first hours postfertilization when the experiments described here were performed and might be higher at later stages of development. Further analysis of sirtuin activity at different stages of zebrafish development will be needed for optimization of SIRTify. Lastly, the serum stability of p86 and derivatives is unknown, and alternate sequences may be required for extended measurements in animal models. As previous studies have demonstrated that most cell-penetrating peptides display homogeneous distribution, we expect similar tissue distribution of the p86 peptide in vivo.⁴¹ Modifications of p86 by the addition of non-natural amino acids or packaging in nanoparticles could be used to improve stability.^61,62

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Michael Deininger (mdeininger@versiti.org).

Materials availability

Plasmids and cell lines generated in this study will be made available on request, but we may require a payment and/or a completed Materials Transfer Agreement if there is potential for commercial application.

Data and code availability

Raw data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Animals

Naive NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl/SzJ (NRG) mice were purchased from the Jackson Laboratory. Female mice aged 8–10 weeks were used for experiments. All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Utah (Salt Lake City, UT). All experiments and husbandry of zebrafish were approved by and conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) at the University of Utah. Adult zebrafish were purchased from and maintained by the Centralized Zebrafish Animal Research (CZAR) at the University of Utah.

Cell lines

All cells were cultured at 37°C in a humidified incubator supplied with 5% CO₂. HEK293T/17 (human embryo) cells were cultured in Dulbecco’s Minimum Essential Medium (DMEM, ThermoFisher) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO) and 1% penicillin/streptomycin (Invitrogen). HepG2 (human male) and KG1 (human male) cells were grown in DMEM supplemented with 10% FBS and 100 U/mL penicillin/streptomycin (P/S). Cells were authenticated using the GenePrint 24 kit (Promega) at the DNA Sequencing Core Facility, University of Utah. All cell lines were screened for mycoplasma using the MycoAlert Mycoplasma Detection Kit (Lonza) and were negative.

METHOD DETAILS

Peptide sequences

All peptides were synthesized, and sequences verified by Vivitide (now Biosynth, Ltd).

p86: H₂N-VSGWRLFKKIS-OH

p86-R4: H2N-VSGWRLFKKISRRRR-OH

p86-Acetyl_8,9: H₂N-VSGWRLF(K_Ac)(K_Ac)IS-OH

p86-Acetyl₈: H₂N-VSGWRLF(K_Ac)KIS-OH

p86-Acetyl₉: H₂N-VSGWRLFK(K_Ac)IS-OH

p86-Crotonyl_8,9: H₂N-VSGWRLF(K_Cro)(K_Cro)IS-OH

p86-Glutaryl_8,9: H₂N-VSGWRLF(K_Glut)(K_Glut)IS-OH

p86-Glutaryl_8,9-R4: H₂N-VSGWRLF(K_Glut)(K_Glut)ISRRRR-OH

p86-Succinyl_8,9: H₂N-VSGWRLF(K_Suc)(K_Suc)IS-OH

p86-Succinyl₈: H₂N-VSGWRLF(K_Suc)KIS-OH

p86-Succinyl₉: H₂N-VSGWRLFK(K_Suc)IS-OH

p86-Succinyl_8,9-R4: H₂N-VSGWRLF(K_Suc)(K_Suc)ISRRRR-OH

Cell-free assays

Assays were set up manually in white flat bottom 96-well plates (BRANDplates; Sigma-Aldrich) at room temperature. All 100 μL reactions were performed in sirtuin buffer, containing 50 mM Tris-HCL, pH 8.0, 137 mM NaCl, 2.7 mM KCl, 1 mM MgCl₂, 1 mg/mL BSA (Enzo Life Sciences). All reaction mixtures contained 250 nM of sirtuin (SIRT1–3, 5 Reaction Biology, SIRT4,6,7 Sigma-Aldrich), 10 μM furimazine (Promega), and 250 μM NAD⁺ (Enzo Life Sciences), unless otherwise stated, and were prepared in sirtuin buffer. Reaction mixtures were incubated at room temperature for 30 min. The reaction was initiated by the addition of 1–250 nM acylated peptide (Vivitide) and a 1:10,000 dilution of LgBiT subunit of the NanoBiT luciferase (Promega) and read using an Envision plate reader (XCite 2105, PerkinElmer).

Kinetic measurements

Assays were performed in white, flat-bottom 96-well plates (BRANDplates; Sigma-Aldrich). Peptides and enzymes were serially diluted in sirtuin buffer for 12–18 concentrations, ranging from X to Y. A mixture of furimazine (10 μM) and LgBiT (1:10,000) in sirtuin buffer was added quickly into wells and briefly mixed before luminescence was measured at 5–10 s increments for 10–15 min at room temperature. Kinetic variables were determined by fitting the hyperbolic curve equations below, where RLU_max is maximum luminescence and S is the NanoLuc substrate.

$$RLU=\frac{RL{U}_{\mathit{max}}\times \left[\mathit{peptide}\right]}{K_D+\left[\mathit{peptide}\right]};RL{U}_{\mathit{max}}=\frac{V_{\mathit{max}}\times \left[S\right]}{K_M+\left[S\right]}$$

For sirtuin kinetic measurements, acylated peptide was added at 250 nM after the addition of furimazine and LgBiT mixture.

Lentivirus production

Plasmids were transfected as a stoichiometric mixture (21 μg) in HEK293T/17 cells using Lipofectamine 2000 and Plus Reagent (Invitrogen) together with psPAX2 was a gift from Didier Trono (15 μg) (Addgene plasmid #12260; http://n2t.net/addgene:12260; RRID:Addgene_12260) and pVSV-g (10 μg) (Addgene plasmid #132776; http://n2t.net/addgene:132776; RRID:Addgene_132776)⁶³ to generate lentiviral particles. The virus was concentrated with PEG and stored at −80°C.

NanoLuc expression in HepG2 and KG1α cells

One million HepG2 and KG1α cells were plated in standard medium in the presence of polybrene (8 μg/mL) and transfected with a pCDH-CMV-LgBiT-EF1-TagRFP or pCDH-CMV-MTS:LgBiT: GFP-EF1-TagRFP plasmid. Four days after infection, cells were sorted for RFP or RFP/GFP expression and expanded in culture for one week. Luminescence was then measured by adding 10 μM p86-R4 to 50,000 cells in the presence of furimazine and measured using Envision plate reader.

Fluorescence staining

HepG2-LgBiT expressing cells were grown in regular media in 8-well chamber slides (ThermoScientific). Cells were washed with PBS, then stained with prewarmed (37°C) DMEM without phenol red supplemented with 10% FBS and 1% P/S containing MitoTracker Deep Red probe (ThermoFisher) for 30 min at 37°C. Cells were washed with PBS before staining with Hoechst 33342 (ThermoFisher) for 10 min at room temperature. Cells were maintained in DMEM solution during imaging.

Generation of SIRT5 overexpression cells

One million HepG2-LgBiT or KG1α-LgBiT cells were plated in standard medium in the presence of polybrene (8 μg/mL) and transfected with pCDH-CMV-SIRT5-FLAG-EF1-CopGFP.⁶⁴ Cells were then processed as described above. Cells obtained in this manner were analyzed for SIRT5 expression by immunoblot.

Inhibition and activation of sirtuins

Cell-free assays were performed in white, flat-bottom 96-well plates (BRANDplates; Sigma-Aldrich) at room temperature. Inhibitors (SIRT1; Selisistat (EX527), SIRT2; SirReal2, SIRT3; 3-TYP, SIRT5; SIRT5 Inhibitor 1) were combined with sirtuin, and NAD⁺, in sirtuin buffer for 30 min at room temperature while rocking. Peptide was then added to each well and incubated for 10–30 min. Furimazine was added to wells, gently mixed, and read.

For cell-based assays, cells were plated at 50,000 cells per well and allowed to adhere or settle overnight. Cells were washed with warmed PBS and then treated with an inhibitor or activator (see figure for concentration) in complete DMEM media for 24 h at 37°C. After incubation, cells were washed and then treated with peptide in NP-40 and Halt^™ Protease and Phosphatase Inhibitor Cocktail (ThermoFisher) for 10 min at room temperature while rocking. Furimazine was added directly before the plate was read. For intact cells, after incubation, cells were washed and then treated with a peptide in DMEM media without phenol red for 2–4 h at 37°C. Furimazine in DMEM without FBS was added to cells for 5 min before the plate was read. All samples were normalized to p86 peptide.

Bioluminescence imaging of SIRT5 activity in cells

The assay was performed manually in black, flat bottom 96-well plates (BRANDplates; Sigma-Aldrich). Cells were plated at 50,000 cells per well and allowed to adhere or settle overnight. Cells were washed with warmed PBS before the addition of peptide in DMEM (without phenol red) with 10% FBS (ThermoFisher) for 2–4 h at 37°C. Furimazine was added 5 min before the plate was read.

Bioluminescence imaging in vivo

For imaging in mice, unmodified parental HepG2 or HepG2-LgBiT cells were injected into the left/right flank of NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl/SzJ (NRG) mice (Jackson Laboratory, 00779). Once tumors reached ≥5 mm, a mixture of p86-R4 or p86-Succinyl8,9-R4 and furamizine in a PEG-300 solution (10% glycerol, 10% ethanol, 10% hydroxyproplycyclodextrin, 35% PEG-300 in water) was injected intratumorally. The surface of the skin was wiped after injections. Mice were injected 2–3 min apart. Imaging began immediately post-injection. Images were collected every minute under (Low sensitivity settings) Emission filter, open; field of view, 25 cm; f-stop 8; binning, 1 × 1; and exposure time, 1 s. (High sensitivity settings) Emission filter, open; field of view, 25 cm; f-stop 1.2; binning, 2 × 2 and exposure time, 60 s. Exposure times averaged 1 min for 15–30 min. Imaging was performed using IVIS 200 Spectrum, and analysis was performed with Living Image software (Revvity).

Bioluminescence detection in zebrafish

Embryos were obtained from crosses of wildtype adult TuAB strain zebrafish (Danio rerio) and injected at the 1-cell stage. For preliminary testing of endurazine (Promega, N2570) and furimazine (Promega, N1120), 5 μL reactions were made by adding 4 μL LgBiT (Promega, N1120), 0.5 μL 100 μM p86-R4 (Vivitide) or 0.5 μL water, and 0.5 μL (10% of reaction volume) 0.5% Phenol Red dye. 100 embryos per treatment group were injected into the yolk with ~1.5 nL per embryo of the above reactions, and 100 uninjected siblings were set aside for “no luciferase” controls. 20 embryos from each group were plated into a single well of a white, flat bottom 96-well plates (Bandplates; Sigma-Aldrich) at room temperature in 200 μL 20 mM HEPES buffered E3 media (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgCl2), pH 7.5, supplemented with 10% FBS with endurazine (1:100) or furimazine (1:50). Following a 30-min incubation period luminescence was measured using a TECAN M1000 with kinetic cycle of 1-min interval for a total of 90 min. Experiments were performed in triplicate. After establishing endurazine as the appropriate substrate for this application, the above experiment was repeated, only including a reaction containing 10 μM p86-Succinyl_8,9-R4 to measure SIRT5 activity in early-developing embryos.

Immunoblot analysis

For immunoblotting, cells were lysed in 1× RIPA lysis buffer (ThermoFisher) containing Halt Protease and Phosphatase Inhibitor Cocktail (ThermoFisher). Protein concentration was measured using Pierce BCA Protein Assay Kit (ThermoFisher). Cellular lysates were boiled in Laemmli sample buffer for 10 min, separated on Tris-glycine/SDS-PAGE gels (Bio-Rad), followed by transfer to 0.45 μm nitrocellulose membranes (Bio-Rad). Membranes were blocked in 5% non-fat milk in TBST buffer for 1 h at room temperature, then incubated with primary antibodies for 2 h at room temperature or overnight at 4°C with gentle rocking. Rabbit monoclonal anti-SIRT5 (D5E11, Cell Signaling) and rabbit monoclonal anti-β-actin (13E5, Cell Signaling) antibodies were used at concentrations of 1:1000 and 1:2000, respectively. Membranes were washed three times for 5 min before secondary antibody was added for 1 h at room temperature with gentle rocking. Secondary antibodies used were IRDye 800CW anti-rabbit (926–32213, LI-COR). Membranes were washed three times before being imaged with an Odyssey Fluorescent Imaging System (LI-COR, Lincoln, NE). ImageJ software was used to analyze the optical density quantification of immunoblots.⁶⁵

Statistical analysis

All experiments were performed in triplicate, independently. Prism 9 (GraphPad) was used to perform all statistical analyses. Please see the figure legend for the statistical analysis performed. p < 0.05 was considered to be statistically significant, except where noted. Multivariant studies were analyzed and p values were calculated using one-way ANOVA, or two-way ANOVA with Tukey’s method of adjustment for multiple comparisons.

Supplementary Material

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.chembiol.2024.10.006.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
IRDye® 800CW anti-rabbit	LI-COR Biosciences	Cat# 926–32213; RRID: AB_621848
Chemicals, peptides, and recombinant proteins
psPAX2	psPAX2 was a gift from Didier Trono	RRID: Addgene_12260
pVSV-g	pVSV-G was a gift from Akitsu Hotta (Stewart et al.63)	RRID: Addgene_132776
LgBiT Protein	Promega	Cat# N401C
Nano-Glo® Luciferase Assay Substrate	Promega	Cat# N113A
Nano-Glo® Endurazine™ Substrate	Promega	Cat# N2570
Selisistat	MedChemExpress	Cat# HY-15452
3-TYP	MedChemExpress	Cat# HY-108331
SIRT5 Inhibitor 1	MedChemExpress	Cat# HY-112634
SirReal2	MedChemExpress	Cat# HY-100591
p86-R4: H2N-VSGWRLFKKISRRRR-OH	Vivitide	N/A
p86-Acetyl8,9: H2N-VSGWRLF(KAc)(KAc)IS-OH	Vivitide	N/A
p86-Acetyl8: H2N-VSGWRLF(KAc)KIS-OH	Vivitide	N/A
p86-Acetyl9: H2N-VSGWRLFK(KAc)IS-OH	Vivitide	N/A
p86-Crotonyl8,9: H2N-VSGWRLF(KCro)(KCro)IS-OH	Vivitide	N/A
p86-Glutaryl8,9: H2N-VSGWRLF(KGlut)(KGlut)IS-OH	Vivitide	N/A
p86-Glutaryl8,9-R4: H2N-VSGWRLF(KGlut)(KGlut)ISRRRR-OH	Vivitide	N/A
p86-Succinyl8,9: H2N-VSGWRLF(KSuc)(KSuc)IS-OH	Vivitide	N/A
p86-Succinyl8: H2N-VSGWRLF(KSuc)KIS-OH	Vivitide	N/A
p86-Succinyl9: H2N-VSGWRLFK(KSuc)IS-OH	Vivitide	N/A
p86-Succinyl8,9-R4: H2N-VSGWRLF(KSuc)(KSuc)ISRRRR-OHSirReal2	Vivitide	N/A
Experimental models: Cell lines
HepG2	ATCC	Cat# HB-8065, RRID: CVCL_0027
HEK293T/17	ATCC	Cat# CRL-11268, RRID: CVCL_1926
KG1	ATCC	Cat# CCL-246, RRID: CVCL_0374
Experimental models: Organisms/strains
NOD rag gamma (NRG) mouse (JAX™)	Jackson Laboratory	RRID: IMSR_JAX:007799

Highlights.

• SIRTify offers higher sensitivity and overcomes limitations of traditional sirtuin assays

• Extends detectable sirtuin activities to include decrotonylase and deglutarylase

• SIRTify is a homogenous assay, enhancing specificity and reproducibility

• Potential for broader applications in measuring additional sirtuin modifications

SIGNIFICANCE.

The split-luciferase approach, SIRTify, significantly advances the measurement of sirtuin activity by providing higher sensitivity and overcoming other limitations of commonly used sirtuin activity assays. Specifically, in contrast to FLUOR DE LYS, which may cause artifacts due to trypsin-mediated degradation in the development step, SIRTify is a homogenous assay, which eliminates the need for a second step, thereby enhancing specificity and reproducibility. Furthermore, SIRTify extends the range of detectable sirtuin activities beyond deacetylation and desuccinylation to include decrotonylase and deglutarylase activities, offering a more comprehensive and quantitative assessment of substrate specificity. The approach holds potential for broader applications in measuring additional sirtuin modifications.