Identification of SIRT3 as an eraser of H4K16la

Summary Lysine lactylation (Kla) is a novel histone post-translational modification discovered in late 2019. Later, HDAC1-3, were identified as the robust Kla erasers. While the Sirtuin family proteins showed weak eraser activities toward Kla, as reported. However, the catalytic mechanisms and physiological functions of HDACs and Sirtuins are not identical. In this study, we observed that SIRT3 exhibits a higher eraser activity against the H4K16la site than the other human Sirtuins. Crystal structures revealed the detailed binding mechanisms between lactyl-lysine peptides and SIRT3. Furthermore, a chemical probe, p-H4K16laAlk, was developed to capture potential Kla erasers from cell lysates. SIRT3 was captured by this probe and detected via proteomic analysis. And another chemical probe, p-H4K16la-NBD, was developed to detect the eraser-Kla delactylation processes directly via fluorescence indication. Our findings and chemical probes provide new directions for further investigating Kla and its roles in gene transcription regulation.


INTRODUCTION
The fundamental subunit of eukaryotic chromatin is the nucleosome.It is composed of an octamer of 8 histone proteins, including 2 copies of H2A, H2B, H3, and H4, wrapped around by a 147 base-pair duplex DNA.The highly packaged and condensed chromatin formed by repeated nucleosome units represses transcriptional activity. 1 However, histone post-translational modifications (PTMs) reveal new insights into gene transcription regulation.With the development of mass spectrometry-based proteomic analysis, numerous types and sites of histone PTMs have been discovered, mainly occurring at the N-and C-terminal tail of histones.Dynamic changes in histone modifications create temporary opportunities for nucleosome-DNA unwrapping and rewrapping, exposing specific sections of DNA for protein binding, resulting in the activation or suppression of gene transcription. 2,3actate, the end product of anaerobic glycolysis, replenishes 2 ATP per glucose without oxygen consumption.In 2019, it was discovered that lactate is not only a metabolite but also a signaling molecule.Zhang et al. 4 reported lysine lactylation as a novel histone PTM that directly stimulates gene transcription similar to histone acetylation.Mass spectrometry analysis identified 26 histone lactyl-lysine sites in human HeLa cells.Additionally, 13 C isotopic labeling techniques confirmed that lactylation levels are highly sensitive to lactate production from glycolysis.Notably, the accumulation of lactylation at the H3K18 site acts as a ''lactate timer,'' stimulating the transcription of M2-macrophage polarization genes and activating M2-related wound healing pathways to regulate macrophage polarization.The discovery of histone lactylation is a significant milestone toward understanding histone dynamics.
Histone deacetylases (HDACs), including HDACs and Sirtuin family proteins, are well studied PTM erasers that regulate gene transcription.Recently, Moreno-Yruela et al. discovered that class I HDACs (HDAC1-3), especially HDAC3, are effective lactylation erasers. 5They found that the in vitro enzymatic delactylation activity of HDAC1-3 is higher than that of Sirtuin family proteins.The authors also treated cells with HDAC inhibitors (butyrate, trichostatin A, and apicidin) and a Sirtuin pan-inhibitor (nicotinamide) to compare the intracellular delactylation activities between HDACs and Sirtuins.The results showed that class I HDACs were the major cellular delactylases because the nicotinamide treatment was unable to induce Sirtuin-regulated pan-Kla increasement, however, the pan-Kac level was also unchanged after treatment with nicotinamide. 5The catalytic mechanisms and physiological functions of HDACs and Sirtuins are different: HDACs require tandem histidine residues to serve as general base catalysts for deprotonating the zinc-bond water to activate it to attack the carbonyl group of acyl-lysine 6 ; on the other hand, Sirtuins catalyze deacylation reactions but require NAD + (Nicotinamide adenine dinucleotide) molecules for the reaction.Sirtuins highly sense changes in intracellular NAD + levels and regulate gene transcription by modulating protein acylation. 7merging biological and structural evidence suggests that Sirtuins are not limited to erasing acetyl groups from lysine residues but can also remove other forms of acyl groups.Sirtuins can erase a range of acyl-lysine groups, from short acylation with a 2-carbon backbone to long-chain fatty acylation with a 16-carbon backbone. 8The zinc-binding domain and Rossmann fold domain make up the highly conserved catalytic core of Sirtuins. 9,10During the enzymatic reaction, the acyl-lysine first binds to the active site of the Sirtuin enzyme, followed by the binding of NAD + .This binding triggers a series of nucleophilic substitution reactions that catalyze the hydrolyzation of the modification and the formation of a nicotinamide molecule and O'acyl-ADP-ribose. 11,124][15][16] In comparison, enzymatic studies demonstrated that SIRT4 acts as a cellular lipoamidase. 17The human SIRT4 structure hasn't been solved yet, but a X. tropicalis SIRT4 structure revealed the dehydroxymethylglutarylation activity of SIRT4. 18Whereas SIRT5 binds strongly to modifications containing the carboxyl tail, such as succinyl-lysine (Ksuc) and glutaryl-lysine (Kglu). 19,20SIRT6 functions best against long-chain fatty acyl-lysine, such as myristoyl-lysine (Kmyr) and palmitoyl-lysine (Kpalm). 21Conversely, aside from the de-fatty-acyl-lysine function, SIRT3 shows a higher affinity for removing modifications that include hydrophilic hydroxyl-motifs, such as b-hydroxybutyryl-lysine (Kbhb) and 2-hydroxyisobutyryl-lysine (Khib). 22Since lactyl-lysine also has similar hydroxyl-motif, we hypothesize that SIRT3 may potentially display higher delactylation activity compared to other human Sirtuins.
To determine the delactylation activity of Sirtuins, we conducted various biochemical and cellular experiments.Our findings indicate that SIRT3 exhibits higher delactylation activity compared to other human Sirtuins, especially on H4K16 site.Crystal structures showed that SIRT3 interacts with Kla-peptides through a preconfigured hydrophobic pocket that accommodates the hydrocarbon portion of the lactyl-lysine.The interaction between SIRT3 and the hydrophilic part of the lactyl motif is facilitated by a hydrogen bond network that incorporates water molecules.Furthermore, we developed a functionalized chemical probe with a clickable photo-cross-linker, p-H4K16laAlk probe, that successfully captured SIRT3 from cell lysates.A fluorogenic probe, p-H4K16la-NBD, was also developed to directly detect Kla-eraser delactylation processes via fluorescence indication.Our findings and newly developed probes support previous Kla eraser studies and open new avenues for further research in the field.

SIRT3 shows higher delactylation activity on H4K16 site compared to other human Sirtuins
With NAD + consumption/cycling assay, lactylated H3K9 (identified in human HeLa cells), H3K14 (identified in mouse BMDM cells), H3K56 (identified in mouse BMDM cells) and H4K16 (identified in human HeLa cells) sites were chosen, and lactyl peptides were purchased for the preliminary screening as these sites are Sirtuins' ''preferred sites.''These four sites have been widely reported that different modifications on these sites can be erased by Sirtuins.And many co-crystal structures of these acyl-lysines and Sirtuins have been solved (Table S1).Due to SIRT4's demethylglutarylase-and deslipoylasehas-f activity and inability to express in soluble form in E. coli, there is a limited yield of SIRT4 protein during purification. 19Therefore, we first prepared all truncated human Sirtuins based on PDB-published Sirtuin-ligand structure sequences except for SIRT4 for the following enzymatic assays and crystallographic study (Figure S1).
In comparison to the negative control group (no peptide), the SIRT5-7 groups exhibited minimal changes in NAD + levels.While the class I Sirtuins, SIRT1-3, as previous reported, showed a certain delactylation activities against lactyl peptides.Among the tested combination of Sirtuins and lactyl peptides, SIRT3 showed a higher delactylation activity on H4K16la and H3K23la peptide.(Figure 1A; Figures S2A and  S2B).Extending the incubation time to overnight (R12 h), both SIRT1-H3K9la and SIRT3-H4K16la consumed much higher level of NAD + (Figure 1B).Next, we performed ITC (Isothermal titration calorimetry) to compare the binding affinity between SIRT1-3 with H3K9la, H3K14la, H3K56la, and H4K16la peptides.Consistent with the NAD + cycling assay result, the binding affinity of SIRT3-H4K16la measured (76.3 mM) is much higher than SIRT1-2 with these lactyl peptides (Figure 1C; Figures S3-S5).
We also performed HPLC-MS analysis to verify this result.Results showed that in comparison with Sirtuins, the reported robust Kla eraser, HDAC3, showed the highest in vitro delactylation activity against H4K16la.While SIRT3 exhibited the highest erased ratio against H4K16la peptides compared with other Sirtuins (Figures 1D and 1E, and the mass results showed in Figure S6).More peptides were erased by SIRT3 after 3 h of incubation, while SIRT1 only showed slight hydrolytic activity against H3K9la and H4K16la.Additionally, those Sirtuins didn't show any hydrolytic activity against H4K12la peptides (Figure S7).Furthermore, we studied the kinetics of SIRT3-H4K16la by measuring the K M and k cat values via HPLC method.The fitted curve calculated K M = 142.66G 2.64 mM, V max = 0.925 G 0.101 mmol/L/min, k cat = 0.08 G 0.008 s À1 , k cat /K M = 5.40*10 2 s À1 M À1 (Figure 1F; Figure S8).Based on these findings, it seems that the in vitro delactylation activity of SIRT3 on the H4K16 site is comparable in magnitude to its de-b-hydroxybutyrylation activity on H4K16, as indicated by a k cat /K M value of 135.09 s À1 M À1 reported by Zhang et al. 22 However, HDAC3 exhibit significantly higher in vitro delactylation activity, as indicated by the k cat /K M value on lactyl peptides (＞4000 s À1 M À1 ). 5,23Collectively, these results verified the delactylation activity of SIRT3 on H4K16 site.

Binding priority of SIRT3 to D-and L-H4K16la
Lactyl-lysine has structural similarities with other acyl-lysines, such as propionyl-lysine, 2-hydroxyisobutyryl-lysine, and b-hydroxybutyryl-lysine, has a unique chiral center at the 3-carbon-site.Compared to the R-configure Kbhb, SIRT3 exhibits higher binding affinity with the S-configure Kbhb.We investigated whether SIRT3's delactylation activity showed differences on lactylated L-and D-lysine peptides by performing the NAD + consumption/cycling assay and ITC experiments.Results showed that SIRT3 preferred hydrolyzing lactylated D-lysine over lactylated L-lysine, as it consumed more NAD + after incubation with the former.The binding affinity of SIRT3 to lactylated D-lysine peptide was also found to be approximately 2-fold higher than that of lactylated L-lysine peptide.(Figures 1G and 1H; Figure S9).These results proved SIRT3 also preferred to hydrolyze lactylated D-lysine other than lactylated L-lysine.
In vertebrate metabolism, L-LDH plays a crucial role in the conversion of lactate from pyruvate, and the lactate produced from glycolysis is almost exclusively in L-configuration. 24D-lactate, on the other hand, can only be absorbed from exogenous sources or be converted by very few D-LDH. 25Although both HDACs and SIRT3 showed a preference for hydrolyzing lactylated D-lysine, further investigation is necessary to fully understand the physiological existence and function of D-lactylation.This may include exploring the biological processes and pathways that are regulated by D-lactylation, as well as identifying the enzymes involved in D-lactylation and their catalytic mechanisms.

SIRT3 knockdown affects H4K16la level in living cells
Then, we investigated if SIRT3 can regulate intracellular lactylation levels.Cellular localizations of seven human Sirtuins are different.SIRT6 and SIRT7 only exist in nuclei.SIRT1 mainly locates at nuclei but also exists in cytoplasm, 26 while SIRT2 is primarily located at the cytoplasm but also exists in nuclei.SIRT3, SIRT4, and SIRT5 are regarded as the mitochondrial Sirtuins. 27However, more evidence (listed in Table S2) shows that SIRT3 also exists in a nuclei and acts as the histone deacylase.
To investigate SIRT3's delactylation activity on physiological nuclear substrates in living cells, we performed a siRNA knockdown assay in living HEK293T cells.Knocking down either SIRT3 or HDAC3 increased cellular H4K16la level (Figure 1I; Figure S10).Moreover, knocking down SIRT3 resulted in a more significant enhancement of H4K16la levels compared to the knockdown of SIRT1/2 (Figure 1J; Figure S11).These data supported the nuclear localization of SIRT3, acting as a nuclear delactylase.

Structural basis of SIRT3 with lysine lactylation peptides
Next, we performed crystallographic studies to gain insight into the molecular basis of SIRT3's interaction with lactyl peptides.Crystals of H3K23la(L)-SIRT3 and H4K16la(L)-SIRT3 complexes diffracted to 2.0 A ˚and 2.5 A ˚resolution, respectively.A structural model from the PDB: 3GLS, was employed as the search model of molecular replacement.The SIRT3-H3K23la(L) and SIRT3-H4K16la(L) complex structures were well-built and refined (data collection and refinement statistics are displayed in Table S3).
Each structure contains 1 SIRT3 molecule, 1 zinc atom, and 1 lactyl peptide in each asymmetric unit.The 2Fo-Fc map clearly captured more than 6 amino acids' electron density (Figure 2A).The catalytic pocket of SIRT3 displayed half-hydrophobic and half-hydrophilic characters.In both structures, F180, I230, I291, and V292 of SIRT3 formed a preconfigured hydrophobic cage that accommodated the hydrocarbon portion of the lactyl group (Figure 2B).Similar to other published SIRT3-acyllysine-peptide complex structures (PDB: 5BWN, SIRT3 with Kmyr peptide, and PDB: 5Z93, SIRT3 with Kbhb peptide, show in Figure S12), the amido linkage in SIRT3-H4K16la(L) was stabilized by V292, H248, and Q228 though direct hydrogen bonds as well as water bridges (Figure 2C).Several water molecules participated in the hydrogen bond network to facilitate the interaction between the hydroxyl group of the Kla and Q228, N229, H248 of SIRT3 in SIRT3-H4K16la(L) structure (Figures 2C and  2D).While in the SIRT3-H3K23la(L) structure, due to the crystallization conditions involving glycerol, two glycerol molecules' hydroxyl groups replaced the water molecule's position and contributed to the formation of the water bridge, resulting in a robust hydrogen bond system (Figure S13).Collectively, these structural data supported SIRT3's ability to accommodate lactyl-lysine and carry out the subsequent NAD +consumption-based catalytic reaction.

Development of the functionalized chemical probes
Chemical affinity-based probes are powerful tools that convert weak protein-ligand interactions into robust covalent chemical linkages, thus enabling the capture of transient interactions.To validate the delactylation activity of endogenous SIRT3 and profile the transient interactions between other potential erasers/readers and Kla substrates, we first generated a lactate mimic with clickable alkynyl group (Figure S14), in which the alkynyl group could be subsequent biorthogonal conjugated to a biotin tag for the following streptavidin pull-down and proteomic analysis.We tried to use this lactate mimic to induce the increasement of pan-alkynyl-lactylation level.It has been reported that the treatment of sodium lactate on cells for 24 h will induce detectable levels of lysine lactylation. 4While our lactate mimic showed unavoidable cytotoxicity against MCF-7 cells and HeLa cells.Therefore, based on our biochemistry data and previous experience with PTM probes, we designed and synthesized a H4K16la based tagging system that covalently links a H4K16la peptide (p-H4K16laAlk) (Figure 3A).The p-H4K16laAlk consists of 3 parts, i.e., ①the affinity-based part: peptide with H4K16Kla sequence and modification for target protein recognition.②A photo-cross-linker (i.e., diazirine) at the N-terminus of the peptide, which could convert the transient interactions between peptide ligand and proteins into robust and irreversible covalent bonds after ultra-violet (UV) irradiation.③ An alkynyl group for biotin clicking.][30] We performed the UV induced photo-cross-linking in 2 groups, group p-H4K16laAlk and group p-H4K16laAlk with competitor (H4K16la peptide) in U251 cell lysates.And after click chemistry reaction, the proteins bound in these two groups were pulled down by streptavidin agarose beads and sent for HPLC-MS/MS analysis.After proteomics analysis, we identified 56 proteins that preferentially enriched in group p-H4K16laAlk compared with the group p-H4K16laAlk with competitor .And SIRT3 was identified successfully with enriched ratio R2 and with p value % 0.05 (Figure 3B; Table S4).The preferentially enrichment of SIRT3 by p-H4K16laAlk was further validated by Western blot assay (Figure 3C; Figure S15).Hence, the probe p-H4K16laAlk validated that H4K16la is the substrate of SIRT3 in cellualr environment.
Additionally, another H4K16la based fluorogenic probe, p-H4K16laNBD, was developed to detect the delactylation process directly.This probe contains the H4K16la sequence and lactylation modification, as well as an O-NBD (Nitrobenzofurazan dye) group that is conjugated to the C-terminus.When the lactyl group is erased by the related enzymes, the exposed amino group can undergo an attack on the O-NBD moiety, resulting in its conversion to an N-NBD group and generating a strong fluorescence that can be easily detected (Figure 3D).Based on the fluorescence generated, we can compare the delactylation activities of various enzymes and calculate the delactylation reaction constant.
The p-H4K16laNBD showed an absorption peak at around 370 nm.After the 10 mM probe was incubated with 0.2 mM SIRT3 and 100 mM NAD + for 2 h, we observed lower peak at 370 nm and a new peak at 482 nm, the generated N-NBD product was verified by LC-MS (Figures S16A and S16B).Further, the fluorescence spectrum was tested under 480 nm excitation (Figure 3E).We used HDAC3 as the positive control.From the results, the probe showed similar fluorescence spectrum with an emission peak at 550 nm after incubating with SIRT3 and NAD + .The other groups showed that sole NAD + or denatured SIRT3 will not affect the fluorescence generating.These results verified the ability of this probe for screening delactylation activity on H4K16 site.Using this probe to screen the delactalytion activities of other Sirtuins (Figure 3F), the results showed that HDAC3 and SIRT3 groups fluoresced at 550 nm while other Sirtuins displayed minimal changes in fluorescence (Figure 3F).We also conduct time-dependent fluorescence measurement to calculate the reaction constant using this probe.And related constants are 0.0102 min À1 for HDAC3, and 0.0059 min À1 for SIRT3 (Figure S16C).Together, these results showed this fluorogenic probe is suitable for the screening of delactylation proteins and measurement of reaction rate for delactylation processes.

DISCUSSION
In this study, our results demonstrate that SIRT3 exhibits the highest activity against the H4K16la site compared to other Sirtuins.Through the complex structures of SIRT3 with lactyl peptides, we have elucidated the binding mechanisms between SIRT3 and lactyl peptides.Our findings reveal that SIRT3 possesses a preconfigured hydrophobic pocket that specifically accommodates the hydrocarbon part of lactyllysine.Furthermore, a water bridge network involving Q228, N229, and H248 of SIRT3 stabilizes the hydroxyl group of lactyl-lysine.Notably, the clickable photo-cross-linker probe, p-H4K16laAlk, together with proteomic analysis, have identified and verified SIRT3 as the binder of the H4K16la substrate.Additionally, the fluorogenic probe, p-H4K16laNBD, was able to detect delactylation activity directly, making it possible to screen the delactylation activities of various enzymes.The findings and newly designed chemical probes of this study could have a significant impact on the field of SIRT3-related Kla regulation, and further research into this area is warranted.

Limitations of study
The catalytic mechanisms of Sirtuins and HDACs differ, implying potential variations in their physiological functions in relation to delactylation activities.In addition to the intracellular knockdown assay performed in this study, it is desirable to conduct experiments such as RNAsequencing and Chip-sequencing to further understand the distinct gene transcription regulation functions of HDACs and Sirtuins.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Figure 2 .
Figure 2. The structural basis of SIRT3 and lactyl-lysine peptides (A) Left: The overall structures of SIRT3(white)-H3K23la(L)(green) complex structure (white) and the ligand models observed in the SIRT3-H3K23la(L) structure.Right: The overall structures of SIRT3(light pink)-H4K16la(L)(cyan) complex structure and the ligand models observed in the SIRT3-H4K16la(L) structure.The 2Fo-Fc maps displayed for the ligands are rendered at 0.9s contour.(B) The preconfigured hydrophobic cage of SIRT3 accommodates the hydrocarbon portion of the lactyl group.Upper: SIRT3(white)-H3K23la(L)(green).Lower: SIRT3(light pink)-H4K16la(L)(cyan). (C) Ligand-protein interaction details in SIRT3(light pink)-H4K16la(L)(cyan) structure.Hydrogen bond is represented as a yellow dashed line, and the water molecule is represented as spheres (marine).The length (A ˚) of the hydrogen bond is labeled next to the dashed line.(D) LIGPLOT diagram list interactions between the H4K16(L) ligand and SIRT3.H4K16la(L) (bond color: light pink) and residues of SIRT3 (bond color: gold) are depicted in ball-and-stick mode.Carbon is represented as a black ball; nitrogen is represented as a blue ball; oxygen is represented as a red ball; water molecule is represented as a marine ball; hydrophobic interactions are shown as red arcs; and hydrogen bond is represented as a green dashed line.The length (A ˚) of the hydrogen bond is labeled next to the dashed line.

Figure 3 .
Figure 3. Development of the chemical probes to investigate lysine lactylation (A) Chemical structure of the p-H4K16laAlk.(B) Proteomic analysis results of p-H4K16laAlk pull down assay (the dotted lines represent p = 0.05 and the enriched ratio = 2).(C) Western blot analysis of detecting the existence of SIRT3 in the pull-down products of group p-H4K16laAlk, group p-H4K16laAlk with competitor and DMSO group (negative control).(D) Chemical structure of the fluorogenic probe, p-H4K16laNBD and the schematic diagram of the reaction mechanism.(E) Fluorescence spectrum of p-H4K16laNBD probe (lex = 480 nm).Probe: 10 mM, SIRT3: 0.2 mM, HDAC3: 0.1 mM, and NAD + : 100 mM.(F) The Fluorescence of p-H4K16laNBD (10 mM) with different enzymes (enzyme concentration: 0.1 mM; lex = 480 nm; lem = 545 nm).Enzymatic reaction condition: enzymes (0.1 mM) in 20 mM HEPES buffer (pH = 8.0) containing 100 mM NAD + at 37 C for 2 h.(No NAD + was added in HDAC3 group).

TABLE d
RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS B Cell lines B Bacterial strains for protein expression d METHOD DETAILS B Preparation of recombinant human Sirtuins B NAD + consumption/cycling assay B Binding affinity measurement B LC-MS analysis for Sirtuins and lactyl peptides B Kinetic study for SIRT3-H4K16la