Tandem Allosteric Effects of Reactant and Product that Promote Deacetylation Cycles in Sir2

Abstract

Post-translational modifications play important roles in the regulation of protein function, with N^ε acetylation being a key reversible modification affecting processes such as transcription, metabolism, and stress responses. Sirtuins, particularly SIRT1 and its ortholog, Sir2, are NAD+-dependent deacetylases that target both histone and nonhistone proteins, including the tumor suppressor p53. Acetylation of p53 on K382 influences its degradation and transcriptional activity. Despite structural studies of the Sir2/acetylated p53 complex, the role of the conserved cofactor-binding loop (CBL) in regulating NAD+ binding and deacetylation remains unclear. Using both conventional molecular dynamics (MD) and parallel cascade selection MD (PaCS-MD) simulations, we investigated the conformational dynamics of the Sir2/acetylated and nonacetylated p53 complexes, focusing on the conformational changes in the CBL of Sir2 in response to p53 acetylation. We identified open and closed states of the NAD+ binding pocket caused by CBL conformational changes depending on p53 acetylation and deacetylation. The forward allosteric effect of the acetylated p53 binding was found to open the NAD+-binding pocket, which is expected to promote NAD+ binding. In contrast, the binding of nonacetylated p53 is significantly weaker, and the reverse allosteric effect drives the pocket closure. These sequential allosteric effects positively accelerate the reaction cycle, which can be considered a “tandem allostery of the reactant (acetylated p53) and the product (deacetylated p53)”. Combining these simulations with entropy transfer analysis, K382 was found to initiate multiple communication routes through strands β7 and β9, and the FEG loop, ultimately converging on the CBL via the helical small domain, the Rossmann-fold domain, or directly from p53, thereby highlighting the critical role of CBL in NAD+ binding and p53 deacetylation.

1. Introduction

Post-translational modifications (PTMs) function as fundamental molecular switches that regulate protein structure, activity, charge, stability, and interactions in both eukaryotic and prokaryotic organisms. − A diverse range of PTMs have been identified, including phosphorylation, acetylation, and methylation. However, N^ε acetylation of lysine stands out as a common, reversible PTM that plays a role in numerous cellular processes, such as transcription, translation, stress responses, detoxification, and carbohydrate and energy metabolism. − By attaching an acetyl group to the ε-amino group of lysine (N^ζ atom of the residue), the N^ε acetylation modification alters a target protein’s local charge and conformation, thereby modulating its function and interaction network. , Lysine acetylation is regulated by lysine acetyl transferase (KAT) and lysine deacetylase (KDAC) enzymes. − The latter group can be further categorized as Zn²⁺-dependent histone deacetylases and NAD+-dependent sirtuins.

Sirtuins (SIRT1–7 in mammals) are evolutionary orthologs of the yeast Sir2 protein and play essential roles in a wide range of physiological and pathological processes, including aging, energy responses to low calorie availability, and stress resistance, as well as apoptosis and inflammation. − These enzymes localize in different parts of the cell: SIRT1 and SIRT2 in the cytoplasm, SIRT3, SIRT4, and SIRT5 in mitochondria, and SIRT1, 2, 6, and 7 in the nucleus. , SIRT1, the homologue of yeast Sir2, has drawn particular attention for its ability to deacetylate not only histone but also nonhistone substrates, − including the tumor suppressor protein p53. ,

p53 forms tetramers, each consisting of multiple domains, including a disordered N-terminal transactivation domain (p53-TAD), proline-rich region, DNA-binding domain (DBD), tetramerization domain, and disordered C-terminal domain (p53-CTD). In response to various cellular stresses, these domains undergo different forms of PTM, including phosphorylation, ubiquitylation, methylation, and acetylation. − Among the two intrinsically disordered regions in p53, p53-TADs are phosphorylated at serine and threonine residues, whereas the p53-CTD, which contains many positively charged residues, is primarily modified via ubiquitylation, methylation, or acetylation at six lysine residues. Of these six residues, K370, K372, K373, and K382 can be targets of all three modifications. These residues are also considered components of nuclear localization signal (NLS) II and III. , More specifically, the region of p53 around K382 with and without PTMs assumes distinct conformations upon binding to different proteins. For example, unmodified p53-CTD binds to Ca²⁺-loaded S100B­(ββ) with an α-helical form (PDB ID: 1DT7 ), but it does not form any specific secondary structure upon binding to the CDK2/Cyclin A complex (PDB ID: 1H26 ). When acetylated at K381 and methylated at K382, p53-CTD binds to the tandem Tudor domain of 53BP1, taking an α-helical form (PDB ID: 4X34 ). A p53 fragment acetylated at K382 (K^Ac382-p53, hereafter) forms a β-turn-like structure upon binding to the bromodomain of the coactivator CREB-binding protein (CBP) (PDB ID: 1JSP ), but another K^Ac382-p53 also folds into a β-strand (PDB ID: 1MA3 ) upon binding to archaeal Sir2, a homologue of human SIRT1. These examples demonstrate the highly disordered nature of the p53-CTD around K382, as well as its high adaptability to different conformations, with and without PTMs.

Acetylation and deacetylation of p53 at K382 is important for regulating the function of p53 (Figure A). DNA damage or oxidation leads to acetylation of p53 at K382 by p300 or CBP, both of which are members of the p300-CBPD coactivator family. By contrast, SIRT1 deacetylates p53 at K382, thereby down-regulating p53 transcriptional activity, facilitating MDM2-mediated p53 degradation, suppressing p53-dependent apoptosis, and promoting cell survival. − ,, Previous studies demonstrated that K^Ac382-p53 initially binds to Sir2, followed by the binding of nicotine adenine dinucleotide (NAD+) to complete the deacetylation process. ,,, To date, several structures of p53 acetylated at K382 and bound to Sir2 have been resolved, including some structures complexed with the bacterial homologue of human SIRT1 from Thermotoga maritima (Sir2-Tm: 1YC5, 2H2D, 3JR3, and 3PDH), and one structure complexed with the archaeal homologue from Archaeoglobus fulgidus (Sir2-Af2: 1MA3). Notably, these structures were all solved by Wolberger and co-workers, whose efforts have provided critical foundational insights into Sir2-substrate interactions. The structures exhibit a high degree of similarity, and the interactions between Sir2 and acetylated p53 are highly conserved across these complexes ().

1.

K382 acetylation and activation of p53 and the structure of the Sir2-Af2 complex with acetylated p53. (A) Regulation of p53 activation through K382 acetylation and deacetylation. Under conditions of cellular stress, inactive p53 is acetylated at lysine 382 (K382) by the coactivator CBP/p300, leading to p53 activation. Activated p53 then initiates cell-cycle arrest, senescence, or apoptosis. Sir2, facilitated by NAD+, deacetylates K382, returning p53 to its inactive state and suppressing the downstream effects. (B) The color-coded crystallographic structure of Sir2-Af2 (cartoon model) in complex with the K382-acetylated C-terminal fragment of p53 (PDB: 1MA3). NAD+ (yellow licorice model) is shown by aligning the original structure with NAD+-bound Sir2-Af2 (PDB: 1S7G ) to better visualize the NAD+-binding pocket. Color scheme: Rossmann-fold domain (RFD, light blue), zinc-binding domain (purple), zinc (white sphere), helical small domain (pink), β7/β9/FGE loop (cyan), acetylated p53 (orange), K^Ac382 (licorice), and cofactor binding loop (CBL, green).

Sir2 is an NAD+-dependent protein deacetylase that relies on NAD+ as a cosubstrate to catalyze the removal of acetyl groups from acetylated lysine residues on protein substrates. Upon binding to K^Ac382-p53, Sir2 subsequently binds to NAD+. During the deacetylation reaction, NAD+ is cleaved into nicotinamide (NAM) and a reactive ADP-ribose intermediate, which accepts the acetyl group from K382 to form 2′-O-acetyl-ADP-ribose (OAADPr). Sir2 consists of a Rossmann-fold domain (RFD), a helical small domain, and a zinc-binding domain, which are connected by two different loops (Figure B). The FGE loop, which contains the highly conserved FGExL motif, functions as a connector that links the zinc-binding domain and RFD (cyan), creating a large groove in which K^Ac382-p53 binds. K^Ac382-p53 (orange) is sandwiched by two β-strands of Sir2 (β7 and β9) upon binding and forms an antiparallel β-sheet. Notably, another flexible α2-α3 loop region connects the relatively large RFD with the small helical domain. NAD+ is positioned close to this loop, suggesting that it plays a crucial role in NAD+ binding; therefore, the flexible α2-α3 loop is referred to as the cofactor-binding loop (CBL). The NAD+-binding pocket is situated predominantly along the RFD, which constitutes the bottom of the NAD+-binding site, whereas the CBL forms the “ceiling” of the pocket. ,

It is important to note that the CBL is highly conserved among Sir2 family members. The CBL contains a GAGΩSX ₃GIPXFRX_3–4GΩΦ sequence motif, where ‘X’ denotes any residue, and Ω and Φ represent hydrophobic and aromatic residues, respectively. This high degree of conservation underscores the critical role of the CBL in the functions of Sir2 family proteins. The conformations of this loop region differ significantly among the limited existing crystal structures of Sir2 homologues. ,,− However, details regarding the role of the CBL in regulating the NAD+ binding pocket and thereby influencing the deacetylation process remain unclear due to challenges in observing the disordered loop region experimentally, coupled with a lack of convincing structural evidence.

In this study, all-atom molecular dynamics (MD) simulation and parallel cascade selection MD (PaCS-MD) − were used to elucidate the molecular mechanism underlying the conformational changes in the CBL upon p53 binding and the role of these changes in controlling the conformation of the NAD+-binding pocket.

To investigate the mechanism of the conformational change in the CBL induced by the binding of p53 with conventional MD (cMD) and PaCS-MD, three systems were simulated: Sir2 bound to a p53-CTD fragment acetylated at K382 (Sir2/K^Ac382-p53); Sir2 bound to a nonacetylated p53-CTD fragment (Sir2/p53); and apo Sir2. The Sir2/K^Ac382-p53 system represents the state before NAD+ binding, which is necessary for the deacetylation reaction. Sir2/p53 represents the potential state just after deacetylation of K^Ac382 or the state just after binding of nonacetylated p53 to apo Sir2; however, the latter should occur infrequently because the binding of nonacetylated p53 to apo Sir2 is at least 3 orders of magnitude weaker than that of K^Ac382-p53. Using cMD, we identified significant conformational changes in the flexible CBL region that are dependent on the acetylation status of the C-terminal domain of p53. PaCS-MD simulations were carried out to enhance the sampling of conformational changes in the CBL. This approach allowed us to capture rare events that are challenging to observe using conventional MD methods. The surface areas of the NAD+-binding pocket were introduced as the selection feature in the PaCS-MD to investigate the conformational changes in the CBL. Based on the results of those analyses, we constructed a Markov state model (MSM) to determine the free-energy landscape (FEL). Our findings reveal that opening of the NAD+-binding pocket of Sir2 is related to the conformational change that is induced in the CBL following the binding of acetylated p53 at K382; however, Sir2 remains in the closed conformation in the absence of p53 binding, even with the binding of nonacetylated p53. Additional analyses of key interactions and entropy transfer pathways identified via entropy transfer analysis are needed to reveal details of the mechanisms of opening and closure.

2. Methods

2.1. Preparation of Simulation Systems

The crystal structure of Sir2-Af2 bound to K^Ac382-p53 (PDB: 1MA3) was employed as the initial structure of the simulation for the Sir2/K^Ac382-p53 complex. A total of 11 missing residues of Sir2, CBL residues 30–39 (sequence: SGIPTFRGED) and the C-terminal residue of Sir2 (K253), were initially modeled using MODELER 10.4. PDB2PQR − was used to assign the protonation states of residues at pH 7.0, resulting in the following histidine states: HIE17, HID82, HID106, HID118, HID130, HIP183, and HIE380, while all other charged residues were assigned their standard protonation states. The N- and C-termini of K^Ac382-p53 and p53 fragments (residues 379–387) were capped by acetyl and N-methyl groups, respectively. 2-(N-morpholino)-ethanesulfonic acid served as the buffer in the crystallization process and was not included in the simulation systems. The initial structures for the Sir2/p53 complex and apo Sir2 were generated by the replacement of acetylated K382 (K^Ac382 hereafter) with lysine and removal of K^Ac382-p53 from Sir2/K^Ac382-p53, respectively.

All cMD simulations were performed using GROMACS. Parameterizations of the three systems were conducted using LEaP, applying the AMBER ff19SB force field for the proteins, with OPC serving as the water model, with subsequent conversion of the input file format from Amber to GROMACS. The acetylated lysine (residue label: “ALY”) in p53 was parametrized using AMBER ff19SB_modAA. The Zinc Amber Force Field was utilized specifically for zinc and the four coordinating cysteine residues in Sir2. The system box was set to a cube with edges of 12.0 nm, and the complex was neutralized at 0.10 M KCl, with the temperature set at 300 K and pressure at 1 bar. All the simulations were performed under equilibrium conditions without applying any biasing forces or potentials. The detailed cMD protocol is described in Supporting Information (S1).

2.2. cMD Simulation to Characterize the Open and Closed Conformations of Sir2

Preliminary cMD simulations of Sir2/K^Ac382-p53 and Sir2/p53 were conducted for 1 μs. The observed increase in root-mean-square deviations (RMSDs) of the atoms in both systems indicated significant conformational changes had occurred, particularly in the CBL. Subsequently, a clustering analysis was performed on both systems to identify the most representative structures. Surprisingly, these analyses revealed that Sir2/K^Ac382-p53 adopts an open state of the NAD+-binding pocket, whereas Sir2/p53 exhibits a closed state.

To validate the observed tendency of the CBL to undergo conformational changes, we performed more extensive simulations starting from different initial conformations. For each case, five independent 2-μs cMD trials were conducted (Table ). We first carried out five cMD trials of Sir2/p53 starting from the open conformation (Open). This initial structure was obtained as representative of the most populated cluster resulting from the preliminary cMD trajectory of Sir2/K^Ac382-p53 and by replacing K^Ac382 with a nonacetylated lysine. The five trials were conducted independently from the aforementioned equilibration steps with different initial velocities. Clustering analysis of the cMD trials resulted in the discovery of two distinct closed conformations, designated Close1 and Close2. Subsequently, five cMD trials were conducted for Sir2/K^Ac382-p53 from Close1 and Close2 to investigate the potential transition to the open state. The initial Close1 and Close2 structures were obtained as presentative structures of the production cMD trials of Sir2/p53. As expected, conformational transitions from Close1/2 to Open were observed in 7 of 10 trials for Sir2/K^Ac382-p53, whereas the initial states were maintained in two cases. Interestingly, one of the simulations reached an additional closed state that differed significantly from both Close1 and Close2 (designated Close3). A more detailed characterization of Close1–3 is provided in the next section.

1. Open/Closed States of Sir2 Reached by 2 μs cMDs Starting from Different Initial States.

		number of final states
system	initial state	Open	Close1	Close2	Close3
Sir2/KAc382-p53	Close1	4	1	0	0
	Close2	3	0	1	1
Sir2/p53	Open	3	1	1	0
Apo Sir2	Open	3	1	0	1
	Close1	2	1	1	1
	Close2	0	0	3	2

The behavior of apo Sir2 must also be considered, as a previous study showed that p53 binds to Sir2 prior to NAD+ binding. Therefore, we performed cMD simulations of apo Sir2 from Open, Close1, and Close2, each with five trials. The initial structures for Open, Close1, and Close2 were prepared by removing K^Ac382-p53 and p53 from the Sir2/K^Ac382-p53 and Sir2/p53 systems. Although the results suggested a tendency of the CBL to close or open in each system, a more quantitative analysis will be carried out by calculating the conformational change FEL.

2.3. Characterizing the Entrance of the NAD+-Binding Pocket Based on Surface Areas

Quantitative descriptors are necessary for characterizing the closed and open conformations. As opposed to our initial expectation that the NAD+-binding pocket volume is lower in the closed states, volume analyses failed to provide an appropriate quantitative description, as the closed states exhibited significantly larger volumes than the open state (see below). As conformational changes between the open and closed states primarily affect the entrance of the NAD+-binding pocket, we then introduced the arbitrary NAD+-binding pocket entrance surface areas, S1, S1′, and S2 (Figure A). To facilitate this calculation, we identified the key residues that constitute the entrance of the NAD+-binding pocket. By measuring the distance between the C^α atoms of the key residues and their center of mass, it was possible to derive the entrance pocket area by summing the areas of the resulting triangles, thus providing a more insightful characterization of the conformational changes induced by the opening and closing of the CBL. For Sir2/p53 and Sir2/K^Ac382-p53, the C^α atoms on Sir2 and p53 were used to calculate the entrance areas: C^α atoms of A28, S30, G31, A218, and K234 on Sir2, and T387 on p53-CTD for S1; and C^α atoms of E38, D39, G40, L41, A218, E219, and P220 on Sir2 for S2. For apo Sir2, which lacks a comparable p53 C^α atom, a pseudo atom defined as the center of two atoms of the helical small domain (C^α atom of V50) and RFD (C^α atom of V195) was introduced to define a new surface area, S1′, as the equivalent of S1 for apo Sir2. Figure B shows two distinct closed states (Close1 and Close2) identified for Sir2/p53. Close1 is characterized by a small S1 (S1′), whereas a large S2 corresponds to “S1 Closure”. By contrast, Close2 with a small S2 and large S1 (S1′) represents “S2 Closure”. An additional close state found for apo Sir2, Close3, which is characterized by a small S1 (S1′) and S2, can be referred to as “S1/S2 closure”. Changes along S1 (S1′) and S2 delineate distinct opening/closing pathways, highlighting their unique features of conformational changes around the NAD+-binding pocket. As mentioned previously, the volume of the NAD+-binding pocket in the open state is smaller than those of Close1–3 (Figure C). Examples of the S1 and S2 closure processes during 2-μs MD are shown in Movies S1 and S2, respectively. Similarly, Movies S3 and S4 show examples of the S1 and S2 opening processes, which involve conformational changes from Close1 to Open and Close2 to Open, respectively. Figure S3 shows examples of time evolution of surface areas S1/S1′/S2 in cMD trajectories, which indicate that opening and closure occur in a time range of tens to hundreds of nanoseconds.

2.

Surface areas of the entrance of the NAD+-binding pocket, representative structures of the open and closed states, and NAD+-binding pocket volumes. (A) Surface areas of the entrance of the NAD+-binding pocket, S1′, S1, and S2. Surface areas were calculated by connecting each selected atom (shown as a yellow ball with residue indices labeled) with the center (shown as a cyan ball) and summing the areas of the resulting triangles. Pseudo atoms (purple balls) were defined as the center of two atoms from the helical small domain and RFD, respectively, to facilitate the calculation of S1′. (B) Different states of the NAD+-binding pocket, Open and Close1–3. Orange and green regions represent p53 and CBL, respectively. NAD+ (spheres), not included in the system, was superimposed to illustrate the binding position of NAD+ in the pocket. (C) NAD+-binding pocket volumes, calculated using PyVOL, are shown in gray for the open and closed conformations. The open state exhibited the lowest pocket volume, at 1639 ų, whereas the closed states showed larger volumes, at 2760 ų (Close1), 1928 ų (Close2), and 2131 ų(Close3). Nicotinamide (NAM) is highlighted in cyan, and ADP-ribose is shown in yellow. The NAD+-binding site comprises three subpockets: A, B, and C, with NAM occupying subpocket C and ADP-ribose spanning subpockets A and B.

2.4. Procedure of PaCS-MD Simulation

cMD simulations are often limited to sampling conformations around specific energy minima, making it challenging to overcome high-energy barriers within a practical computational time frame. To address these limitations, advanced sampling techniques such as PaCS-MD have exhibited strong performance for efficient sampling of large protein conformational changes. ,,,

In PaCS-MD, n _rep (number of replicas simulated in each cycle) replicas of short MD simulations are performed, followed by ranking of the selection feature in this cycle. The selected top-ranked structures based on “selection feature” from the n _rep replicas are then used as initial structures for the next cycle of parallel short MD simulations. All the MD simulations in PaCS-MD were conducted with the isothermal–isobaric condition at 300 K and 1 bar, without applying any biasing forces or potentials. The iterations of short MD simulations and ranking continue until the selected features reach a predetermined threshold value. The selection protocol considerably increases the probability of relatively rare events occurring in each cycle, thus greatly enhancing conformational sampling along the selection feature after the accumulation of many cycles. Convergence of sampled areas was assessed using a bootstrap approach. Further details are provided in the Supporting Information (Figure S1). The results indicate that the distributions were well converged within the number of trials conducted. PaCS-Toolkit, the optimized software utilized for PaCS-MD, is highly adaptable, user-friendly, and can adopt various selection features to suit different research needs. In addition, when combined with the MSM, PaCS-MD/MSM accurately estimates the FEL, thereby enabling the prediction of stable conformations. ,,,,

cMD simulations alone are insufficient for effectively sampling conformational changes in the CBL and confirming the tendencies of opening/closing in each system. To enhance sampling of conformational changes around the CBL, key values identified from the cMD results (S1, S1′, and S2) were utilized as the selection features for PaCS-MD trials. The sampling consisted of two rounds of PaCS-MD simulations (Table ). The first round started from the conformations representing Open, Close1, or Close2, selected from the cMD trials (Table ), and samples toward opposite states (Open → Close1, Open → Close2, Close1 → Open, or Close2 → Open). Sampling of the second round started from the reached state to the original state. As this version of PaCS-MD was conducted using surface areas (sa) as the selection feature, it is designated saPaC-MD hereafter. Open to Close1 and Close2 simulations were conducted by selecting the snapshots with the least S1 (or S1′) and S2 values, respectively. Conversely, selection of snapshots with the largest S1 (or S1′) and S2 was employed for sampling from Close1 and Close2 to Open, respectively.

2. Settings for the Surface Area PaCS-MD (saPaCS-MD).

system	selection feature	n rep	n trial	first round	second round
Sir2/KAc382-p53	S1	50	20	Open → Close1	Close1 → Open
	S2	50	20	Open → Close2	Close2 → Open
Sir2/p53	S1	50	20	Close1 → Open	Open → Close1
	S2	50	20	Close2 → Open	Open → Close2
Apo Sir2	S1′	50	10	Open → Close1	Close1 → Open
	S2	50	10	Open → Close2	Close2 → Open

The saPaCS-MD protocol involved an initial 1-ns preliminary cMD simulation around each initial conformation (cycle 0), wherein the top 50 selected structures based on selection features were assigned as initial structures for cycle 1. Subsequent saPaCS-MD cycles consisted of 50 replicas (n _rep = 50) of 100-ps cMD simulations, with the top 50 structures from each cycle assigned as the initial structures for the subsequent cycle. The number of saPaCS-MD trials (n _trial) was 20 for Sir2/K^Ac382-p53 and Sir2/p53 and 10 for apo Sir2 in each case.

Additionally, five dissociation PaCS-MD (dPaCS-MD) trials (n _trial = 5) were conducted for both Sir2/K^Ac382-p53 and Sir2/p53 to explore their dissociation pathways and calculate the standard binding free energies. The dPaCS-MD protocol followed the same procedure as saPaCS-MD, but 64 replicas per cycle were used (n _rep = 64), and the center-of-mass distance between K^Ac382-p53 (or p53) and Sir2 was employed as the selection feature.

2.5. Principal Component Analysis of Sampled Conformational Spaces

Each saPaCS-MD trial generated a set of unbiased trajectories that exhibited significant overlap in the conformational space due to the short MD simulations starting from snapshots of the previous cycle. Furthermore, given the utilization of different selection features (S1/S1′/S2) in saPaCS-MD simulations, it was essential to assess whether trajectories resulting from different selection directions exhibited sufficiently high overlap in the conformational space. To evaluate this overlap, we employed principal component analysis (PCA), a dimensionality reduction technique that enables the transformation of high-dimensional data sets into lower-dimensional representations while preserving much of the original information. PCA is considered a very useful tool for the sampling of protein conformational spaces using molecular simulations. Principal components capture the maximum variance in the data, with PC1 representing the direction along which the data are the most spread out in variance and PC2 representing the direction orthogonal to PC1, along which the data exhibit the next highest variance. Consequently, we merged all of the trajectories generated by the PaCS-MD simulations (Table ) and conducted PCA of the coordinates of all C^α atoms of Sir2 generated by the PaCS-MD trials. The PCA results revealed significant overlap between the trajectories obtained for Sir2/K^Ac382-p53, Sir2/p53, and apo Sir2 (Figure S4). Correlation coefficients between PCs and S1/S1′/S2 indicate that S1 and S1′ show the strongest correlation with PC3 and that S2 exhibits the strongest (negative) correlation with PC1 (Figure S5), but the surface area projections show better separation of the conformational states as shown in Results.

2.6. Markov State Modeling, Free Energy Landscape of Sir2 Conformational Changes, and Binding Free Energy Calculations for Sir2/K^Ac382-p53 and Sir2/p53

The FEL of the Sir2 conformational change was calculated using the MSM constructed from the trajectories sampled using saPaCS-MD simulations. The MSM was calculated using S1′and S2 as the MSM features. To construct the MSM, we utilized PyEMMA2.5.12, which was used to discretize the conformational space into a set of microstates based on the values of S1′ and S2 using k-means clustering (1000 clusters) with k-means++. These microstates were used to construct a transition probability matrix (M) for which the element M _ij represents the transition probability from microstate i to j. The FEL is then obtained by projecting the free-energy values of each microstate onto a two-dimensional space spanned by S1′ and S2. The details for the free energy calculation are provided in Supporting Information (S2). A lag time of 40 ps, which provided a flat implied time scale versus lag time relationship (Figure S6), was employed to construct the MSM.

The free energy profiles of the Sir2/K^Ac382-p53 and Sir2/p53 complexes were also calculated using the same procedures as a function of the center-of-mass distance between K^Ac382-p53 (or p53) and Sir2, except that the center-of-mass distance d was used as the MSM feature. MSMs were built using 20 clusters and a lag time of 30 ps (Figure S7). The standard binding free energies ΔG° of the Sir2/K^Ac382-p53 and Sir2/p53 were calculated as the sum of the free energy difference −ΔW between the bound (d < 2 nm) and unbound (3 < d < 5 nm) states and the volume correction term ΔG _v . −

2.7. Analysis of Information Transfer from p53 to Sir2

Entropy is considered as a source of information transfer, which is used not only in information theory but has also been successfully employed to analyze allosteric communication in proteins, where Hacisuleyman and Erman introduced a method to quantify entropy transfer between residues. The entropy transfer from residue i to j is denoted as T _ij , and the net entropy transfer from residue i to j is introduced as ΔT _{i →j} = T _{i →j} − T _{j →i} . Detailed calculation of entropy transfer can be found in Supporting Information (S3).

The cMD trajectories of Sir2/K^Ac382-p53 from Close1 and Close2 to Open and those of Sir2/p53 from Open to Close1 and Close2 were subjected to net entropy transfer analysis, after removing the transient parts at the beginning and end so that only equilibrium segments were analyzed. Analysis of Sir2/K^Ac382-p53 would be expected to provide information regarding the mechanisms of S1 and S2 opening upon K^Ac382-p53 binding, whereas the latter could provide insights into S1 and S2 closures after deacetylation of p53. In this work, the C^α atom was considered representative of each residue, and the time delay value τ was determined to be 5 ns, following the original work. Two key criteria were applied to identify the allosteric communication pathways from K382 to the CBL. First, a positive net entropy transfer between residues i and j, and second, a minimum distance between residues i and j of no greater than 4 Å. Under these conditions, a direct communication pathway is considered to exist between these residues. Starting from K382, we identified surrounding residues within 4 Å and traced these direct pathways iteratively until they reached the CBL.

3. Results

3.1. Binding Affinity of Sir2/K^Ac382-p53 and Sir2/p53

Five dPaCS-MD trials were performed for both Sir2/K^Ac382-p53 and Sir2/p53. The MSM was constructed using the center-of-mass distance between K^Ac382-p53 (or p53) and Sir2 as the input feature to estimate the binding free energies (Table ). Free energy changes as a function of the center-of-mass distance d are shown in Figure S8. The calculated standard binding free energies ΔG° were −7.6 ± 1.5 kcal/mol for Sir2/K^Ac382-p53 and −4.1 ± 0.9 kcal/mol for Sir2/p53. These values were consistent with the experimental values for Sir2-Tm measured by isothermal titration calorimetry (ITC), which are shown as dissociation constants (K _d). The experimental K _d values were also converted to ΔG°. These results clearly indicate that the binding between Sir2 and K^Ac382-p53 is stable enough (K _d is in the μM range). However, the binding of nonacetylated p53 with Sir2 is weaker by 2 orders of magnitude in K _d.

3. Results of Binding Affinity Analysis of Sir2/K^Ac382-p53 and Sir2/p53.

system	dPaCS-MD/MSM (Sir2-Af2) ΔG° (kcal/mol)/K d (μM)	ITC (Sir2-Tm) ΔG° (kcal/mol)/K d (μM)
Sir2/KAc382-p53	–7.6 ± 1.5/2.7	–7.3/4.3 ± 0.5
Sir2/p53	–4.1 ± 0.9/9.9 × 102	>−4.1/> 1000

3.2. Free Energy Landscapes of Sir2

The 2D FEL mapped onto the space spanned by S1′ and S2 obtained by saPaCS-MD/MSM is shown in Figure . Four free-energy minima were identified for apo Sir2 (Figure A), consistent with Open and Close1–3 observed during the cMD simulations shown in Table and Figure B. This result indicated that these four states coexist in apo Sir2. In particular, Close3 is situated at the deepest minimum, whereas Close2 exists at a slightly shallower but broader minimum. Compared with Close2 and Close3, Close1 and Open are significantly less stable in this order. By contrast, Sir2/K^Ac382-p53 showed only one free-energy minimum in the open state, indicating that this state is dominant, which corroborates the cMD simulation results (Table ). The minimum in the open state for Sir2/KAc382-p53 also indicated that the binding of K^Ac382-p53 allosterically induces the opening of the NAD+-binding pocket. By contrast, Sir2/p53 can be trapped primarily in Close1, but it can also be trapped to a lesser degree in Close2 (Figure C), suggesting that even if p53 binding without acetylation at K382 occurs, it cannot induce opening of the NAD+-binding pocket. Therefore, the binding of nonacetylated p53 causes reverse allosteric effects that close the NAD+-binding pocket.

3.

Free energy landscapes of the CBL conformational change in Sir2. The free energy landscapes of the CBL conformational change in Sir2 mapped onto the 2D space spanned by S1′ and S2 for (A) apo Sir2, (B) Sir2/K^Ac382-p53, and (C) Sir2/p53. Four distinct minima Open and Close1–3 and corresponding representative structures are shown by both cartoon and surface representations. NAD+, not included in the simulations, was adopted from NAD+-bound Sir2 (PDB: 1S7G ) and superimposed to show the binding pose by spheres. The CBL is labeled green.

The FEL of apo Sir2 clearly indicated the intrinsic nature of Sir2 to assume the closed forms, but it was found to be relatively flexible compared with the other cases. The binding of K^Ac382-p53 moves the stable state of Sir2 to Open and fixes it there. Without acetylation, Sir2 forms the closed states, but the position of the global free-energy minimum shifts to Close1.

Macrostate assignments were made for both apo Sir2 and the Sir2/p53 complex, and the corresponding transition times between these states were determined (Figure S9). These computed timescales were found to be consistent with those observed in the cMD trajectories.

3.3. Allosteric Communication Pathways for S1 and S2 Opening/Closure

The cMD simulation results and FEL clearly indicated that the binding of K^Ac382-p53 triggers the opening of the CBL (Table and Figure B). By contrast, Sir2 bound to nonacetylated p53 is stabilized in the closed states (Figure C). As K^Ac382/K382 of p53 is located relatively far from the CBL, the mechanism by which this allosteric “signal” is transmitted from p53, especially K^Ac382/K382, to the CBL was investigated by analyzing the net entropy transfer (see Methods).

To investigate the communication pathways from K^Ac382 of p53 to the CBL of Sir2 during the pocket openings from Close1 to Open (S1 opening) and from Close2 to Open (S2 opening) in Sir2/K^Ac382-p53, analysis of net entropy transfer was applied to the four S1 opening and three S2 opening trajectories obtained during the cMD simulation (Table ). In the S1 opening process (Figure A), K^Ac382 initiates two distinct communication pathways with similar magnitudes. One pathway transfers entropy first to the FEG loop and β7 strand of Sir2 and then to the CBL through the helical small domain both directly and indirectly via the zinc-binding domain. The other pathway transfers entropy to the CBL through β9 of Sir2, from which the pathway splits into two routes: one along the RFD and the other through F385 and T387 of p53. In S2 opening (Figure B), K382 transmits signals first to β9, initiating a network of interconnected communication pathways. The major pathway involves direct entropy transfer from β9 to the CBL, whereas another pathway reaches the CBL indirectly via the RFD. The RFD further propagates signals primarily to the CBL but also indirectly via the FEG loop, the zinc-binding domain, and the helical small domain.

4.

Allosteric communication pathways from K^Ac382/K382 to the CBL during the opening and closure processes of the NAD+-binding pocket. (A) S1 opening. (B) S2 opening for Sir2/K^Ac382-p53. (C) S1 closure. (D) S2 closure for Sir2/p53. Each panel depicts the structure-based net entropy transfer pathways and directions between the residues, represented by arrows connecting the C^α atoms. The thickness of the arrows is proportional to the magnitude of the net entropy transfer. Arrow heads and tails are color-coded as follows: p53 (orange); β7/β9 (cyan); RFD (light blue); helical small domain (pink); zinc-binding domain (purple); and CBL (green).

Similarly, the net entropy transfer pathways during closure of the NAD+-binding pocket were elucidated by analyzing one trajectory from Open to Close1 (S1 closure) and another from Open to Close2 (S2 closure) for Sir2/p53 (Table ). In the S1 closure process (Figure C), K382 primarily transfers entropy to FEG loop/β7 directly or in a minor fashion through K381 and H380 along p53 and ultimately to the CBL via the zinc-binding and helical small domains. The minor pathways go directly to β9 or indirectly through K381 and H380 via the RFD before reaching the CBL. In the case of S2 closure (Figure D), entropy transfer from K382 first reaches V196 in β9, which communicates with the RFD, branching into two pathways. In the major pathway, the signal flows to V195 and T197 in β9, and then to M384, F385, and K386 of p53 to the CBL via the helical small domain. The minor pathway from the RFD goes to the CBL via the helical small domain without passing through β9.

In both openings and closings, p53 first transfers entropy to FEG loop/β7 or β9, indicating the importance of signal transduction through the antiparallel β-sheet formed by β7, p53, and β9. Ultimately, the CBL receives entropy from the helical small domain, the RFD, or directly from p53. As seen in the structure-based images shown in Figure , the range of the pathways is relatively confined around p53-CTD, FEG loop/β7/β9, and the CBL for the pocket opening (Figure A and B); however, it spreads out over Sir2 during closure (Figure C and D), which will be revisited in the next section. Collectively, these entropy transfer pathways, originating from K^Ac382/K382 in p53, should illustrate the complexity and coordination of long-range allosteric signaling within the protein. The efficient transfer of these signals across distant domains highlights the intricacy of the mechanisms regulating critical functional changes in the opening and closure of the NAD+-binding pocket.

3.4. Key Interactions that Stabilize the Open and Closed States

Detailed hydrogen bond analyses provided important insights that enhance understanding of the interactions that stabilize the open and closed structures and play roles in allosteric communication. Hydrogen bonds were defined using a donor–acceptor distance cutoff of 0.35 nm and a donor–hydrogen–acceptor angle cutoff of 150°. Salt bridges were determined for positively and negatively charged groups separated by a distance cutoff of 0.4 nm. Interaction occupancies were calculated from all equilibrium conformations, and only those with an occupancy greater than 10% are reported (Table S1). In Sir2/K^Ac382-p53, acetylated p53 resides in the large groove between β7 and β9 of Sir2. Key hydrogen bond interactions between the main-chain atoms of p53-CTD and the adjacent β-strands of Sir2 help form a staggered antiparallel β-sheet consisting of three strands (Figure A). These strands form between the following residue pairs, in the order ‘p53 residue’–‘Sir2 residue’: H380: O–L169: HN; K^Ac382: HN–E167: O; K^Ac382: O–G166: HN; L383: HN–Y197: O; and F385: HN–V195: O. The side chain of K^Ac382 is inserted into a hydrophobic tunnel at the interface between the large and small domains Sir2 and notably also forms a hydrogen bond with the carbonyl oxygen of V163 in the FGE loop with HN^ζ of acetylated K382, where the acetyl group attaches. By contrast, this interaction is significantly weakened in Sir2/p53, highlighting the importance of the acetyl group on the binding between Sir2 and p53-CTD. In the open state of Sir2/K^Ac382-p53, the CBL (residues 25–46) is stabilized by intra-Sir2 interactions, which are interactions between the CBL and the helical small domain (D46-E49, D46-V50, and R43-E69), as well as D46-G37, thus stabilizing the C-terminal side of the CBL (Figure B). Further stabilization is derived from interactions between the middle region and more N-terminal side of the CBL (R43-R36 and R43-P33). The apo open state shows similar stabilizing interactions but lacks the D46-G37 and R43-R36 contacts, indicating a less stable conformation compared with the acetylated p53-bound state.

5.

Key interactions that stabilize the open and closed structures. All interactions shown are hydrogen bonds, except those explicitly stated as salt bridges. Key interactions in Open for Sir2/K^Ac382-p53 (A–C). (A) Interactions between p53 (orange) and β7/β9 (cyan) shown in representative structures. (B) Interactions between the CBL (green) and the helical small domain (pink), along with intra-CBL interactions. R43-E69 is a salt bridge. (C) Interactions between the acetyl group oxygen of K^Ac382 and H^δ of H118. Key interactions in Close1 for Sir2/p53 (D, E). (D) Interactions involving the CBL, p53, and the helical small domain, including intra-CBL interactions. (E) Interactions between the CBL and RFD (light blue). Key interactions in Close2 for Sir2/p53 (F, G). (F) Interactions between the CBL and RFD. R36-E38 is a salt bridge. (G) Intra-CBL interactions. (H) Key interactions in Close3 for apo Sir2. Interactions between the CBL and RFD, as well as intra-CBL interactions. R43-E29 is a salt bridge.

In Close1, which is the most stable state for Sir2/p53, there is significant weakening of the main-chain hydrogen bonds between nonacetylated p53 and Sir2 (e.g., L383: HN-Y197: O and F385: HN-V195: O) and the hydrogen bond between K382: HN^ζ-V163: CO that are stable in Sir2/K^Ac382-p53. Additionally, removal of the acetyl group eliminates the H118-K^Ac382 interaction. As a result, the flexibility of p53 increases notably, particularly around K382 and the C-terminal side of p53-CTD, K386, and T387, situated near the CBL. Interactions between the CBL and the helical small domain (Y45-E65) stabilize the C-terminal end of the CBL in the closed form, and the R43-P33 interaction links within the CBL (Figure D). Stabilization also occurs via interactions between the CBL and p53 (R36-T387 and T34-T387) and between the CBL and RFD (I26-N79 and A28-K234) (Figure E), thus anchoring the N-terminal end of the CBL. Close1 for apo Sir2 exhibits similar stabilizing interactions but lacks the p53-CBL contacts (T34-T387 and R36-T387), making Close1 less stable than Close1 in Sir2/p53 (Figure A).

In Close2 for Sir2/p53, hydrogen bonds between p53 and Sir2 are weakened similarly to Close1; however, the interactions between the main-chain atoms of nonacetylated p53 and Sir2 (L383: HN-Y197: O and F385: HN-V195: O) are maintained to a greater degree compared with the more pronounced weakening observed in Close1. The interactions between the CBL and RFD (R36-L194, and R36-P220) function as a “glue” that links the CBL to the RFD (Figure F). Further stabilization of Close2 is derived from internal CBL interactions (E29-T34, P33-R43, F35-R43, and D39-W42) (Figure G). The apo Close2 conformation shows similar stabilizing interactions, indicating that Close2 remains stable even in the apo state of Sir2.

In the apo Close3 state, interactions between the CBL and RFD (S30-G236) and intra-CBL interactions (R43-P33, R43-I32, R43-E29, G31-A28, E29-G25, and S30-I26) stabilize the structure (Figure H).

Contact difference maps in SI further support these findings (Figure S10). In Sir2/K^Ac382-p53 open state, contacts between β9 (G191–V195) and the RFD (A235, N217, P220, T221, M222) are observed but lost in both Close3 (Apo Sir2) and Close2 (Sir2/p53). This loss likely contributes to the structural shift toward closure.

In Sir2/K^Ac382-p53, the formation of strong interactions between p53-CTD and β7/β9, as well as additional interactions with K^Ac382, together with the contacts between β9 and the RFD probably confines the communication pathways from p53-CTD to the CBL of Sir2 to the narrower region, inducing opening of the NAD+-binding pocket. By contrast, weaker interactions in Sir2/p53 spread the communication pathways, eventually leading to closure of the pocket.

3.5. Conservation of Key Residues

Multiple sequence alignment revealed a striking conservation of key residues across archaeal (Sir2-Af2, -Af1), bacterial (Sir2-Tm, HST1), and human (SIRT1–7) Sir2 homologues (Figure S11). Notably, the key residues that stabilize the open and closed structures, A24, S27, P33, R36, D46, V50, L194, and P220, exhibit high levels of conservation, underscoring their common functional significance within the family. Additionally, the key residues along the communication pathways from K382 to the CBL, V163, F165, and G166 on the β7/FGE loop; L194, V196, and P198 on β9; and D46, P47, V50, A51, and F66 on the helical small domain, were also found to be highly conserved.

Importantly, a subset of these residues plays a direct catalytic role in the deacetylation reaction. K^Ac382-p53 is anchored within a conserved hydrophobic tunnel formed by residues such as F165, L169, and V196 and precisely oriented by a hydrogen bond between the N^ζ atom and the carbonyl oxygen of V163. The reaction is catalyzed by an invariant histidine (H118 in Sir2-Af2), which facilitates nucleophilic attack and stabilizes the transition state via the formation of a transient O-alkylamidate intermediate. The FGE loop residues G166 and E167 further contribute to catalysis by forming backbone hydrogen bonds with the substrate, thereby stabilizing the substrate’s position in the active site. These catalytic residues, together with those governing conformational transitions, constitute a highly conserved structural and functional framework essential for enzymatic activity.

The results of this analysis strongly suggest that the opening/closure mechanisms we elucidated are universal among most proteins of the Sir2 family. These findings promise to significantly enhance our understanding of the pivotal roles played by K^Ac382-p53 and nonacetylated p53 in determining the conformation of the CBL, for which the directions of the allosteric effects are reversed depending on the presence/absence of acetylation, but both K^Ac382-p53 and nonacetylated p53 positively promote the deacetylation cycle. Our work thus offers valuable insights into the fundamental functions of Sir2 proteins across species that contribute to a more complete understanding of their biological functions.

3.6. Tandem Allosteric Effects of K^Ac382-p53 and Nonacetylated p53 on the Deacetylation Cycle

The results of this study demonstrate the mechanism of the efficient deacetylation of K^Ac382-p53, as illustrated by the scheme shown in Figure . First, apo Sir2 exhibits the open and multiple closed states of the NAD+-binding pocket but is mostly stabilized in the closed states (Figure A), which impedes the entry of NAD+ prior to binding of K^Ac382-p53 to Sir2. After binding, the NAD+-binding pocket is stabilized in the open state by the allosteric conformational change in the CBL (Figure B), which facilitates entry of NAD+ to the binding site and the subsequent deacetylation activity on K^Ac382-p53. Even if nonacetylated p53 binds to apo Sir2, p53 should dissociate promptly from Sir2, as the binding of p53 to Sir2 is significantly weaker compared with binding to K^Ac382-p53. Upon deacetylation of the acetylated lysine, NAD+ is broken down into NAM and OAADPr, both of which are expected to be quickly released. The Sir2/K^Ac382-p53/NAD+ complex is reportedly stabilized in the Close3 state (Figure S12), which provides an apparent exit for NAM (Figure S13). By contrast, the structure of the Sir2/p53/OAADPr complex, which was solved for Sir2-Tm bound to 3′-O-acetyl-ADP-ribose (PDB ID: 2H59 ), also adopts the Close3 state (Figure S12). In particular, F33 in Sir2-Tm reportedly undergoes a conformational shift to close the so-called “C-pocket”, preventing NAM rebinding. , After deacetylation, as long as binding is maintained in deacetylated p53, Sir2 tends to close the NAD+-binding pocket via an allosteric effect on the CBL in the opposite direction (Figure C). This reverse allosteric effect might also play a role in inducing the rapid release of NAM and OAADPr, also preventing the reverse reaction. Because NAM is a well-known sirtuin inhibitor (IC50 for SIRT1: 120 μM), mechanisms that push NAM from the pocket would be preferable for a more efficient deacetylation cycle. Overall, we hypothesize that the enzymatic activity of Sir2 as a deacetylase proceeds efficiently according to the following cycle (Figure ): (1) binding of K^Ac382-p53 to apo Sir2; (2) opening of the NAD+-binding pocket due to the allosteric effect on the CBL; (3) binding of NAD+ to Sir2; (4) deacetylation of K^Ac382; (5) release of NAM and OAADPr from Sir2; (6) closure of the NAD+-binding pocket due to the allosteric effect of nonacetylated p53; and (7) release of p53 from Sir2. Before the deacetylation reaction, the forward allosteric effect of the reactant (K^Ac382-p53) induces opening of the cofactor (NAD+)-binding pocket and promotes cofactor binding. After the reaction, the reverse allosteric effect of the product (nonacetylated p53) drives closure of the cofactor-binding pocket, presumably inducing cofactor release and preventing the reverse reaction. Both allosteric effects positively accelerate the reaction cycle, which can be considered a “tandem allostery of reactant and product”. We speculate that similar acceleration mechanisms might be found in other enzymes. In essence, Sir2 demonstrates remarkable adaptability, as transition between different states facilitates selective control of binding pocket accessibility, which is also positively regulated by acetylated and nonacetylated p53.

6.

Tandem allosteric effects of K^Ac382-p53 and nonacetylated p53 on the conformational cycle of Sir2. Apo Sir2 exhibits multiple states but is predominantly stabilized in closed conformations, mostly Close3 and Close2, with the CBL highlighted in light green. Upon binding of the acetylated p53-CTD at K382 (K^Ac382-p53, K^Ac382 highlighted in red and the remainder in yellow), Sir2 transitions to an open conformation (Open), facilitating NAD+ entry and activating deacetylation activity. The ternary complex then shifts to the Close3 state, in which a potential exit for NAM likely enables the release of nicotinamide (PDB ID: 4ZZJ). Following deacetylation, the complex retains 2′-O-acetyl-ADP-ribose (OAADPr) within the active site. The available structure of the Sir2/p53/OAADPr complex (PDB ID: 2H59) corresponds to the Close3 state, in which a conformation shift reportedly prevents nicotinamide rebinding. Dotted arrows indicate steps with insufficient structural evidence, which were not investigated in this work. Finally, the deacetylated p53-CTD (yellow and blue) induces closure of the NAD+-binding pocket, leading to dissociation of the p53-CTD and reversion to the apo form of Sir2.

4. Conclusion and Discussion

Our study employed cMD and saPaCS-MD simulations to investigate the conformational dynamics of Sir2-Af2 bound to p53. In contrast to methods that apply biasing forces or potentials, PaCS-MD preserves the system’s natural dynamics, resulting in more physically interpretable transition pathways. By employing a selection-based cascade strategy instead of introducing artificial biases, PaCS-MD minimizes perturbations to the energy landscape. This is particularly important for accurately capturing intermediate states and kinetic properties, while also avoiding sampling artifacts that can arise from the use of inappropriate or poorly chosen collective variables.

Our study focused on elucidating the allosteric opening and closure mechanisms of the NAD+-binding pocket of Sir2, with the aim of determining how conformational changes in the CBL are affected by the acetylation status of p53. The 2D FELs constructed using saPaCS-MD/MSM revealed distinct stable conformations of the CBL and NAD+-binding pocket. Notably, we observed that the interaction with acetylated p53 tends to keep the pocket open, facilitating NAD+ entry and deacetylation activity, whereas nonacetylated p53 predominantly induces adoption of the closed states, tandemly enhancing Sir2 efficiency upon the deacetylation cycle. The open state of K^Ac382-p53/Sir2 adopts a more flexible and dynamic conformation, which may reflect an overall increase in conformational entropy. Enthalpically, the open state gains some internal hydrogen bonds but slightly loses salt bridges compared to Close2 (Table S2), which might cancel out in total. In contrast, the closed states are more compact and ordered. A comparison of average number of interactions shows that closed structures of p53/Sir2 tend to exhibit a slightly higher average numbers of hydrogen bonds and salt bridges (Table S2). These trends are consistent with the notion that pocket opening may be entropy-favored, whereas closure is likely stabilized by favorable enthalpic contributions.

Previous studies have indicated that the SIRT1/p53 axis holds promise as a target in cancer therapy, with the mainstream focus centered on the development of inhibitors that occupy the NAD+-binding pocket, thereby preventing p53 deacetylation and subsequent degradation. By contrast, our study introduces a potential new approach that targets the closure mechanisms of the NAD+-binding pocket to obstruct NAD+ binding. This insight presents alternative avenues for modulating Sir2 functions that could result in new therapeutic applications.

We elucidated the opening and closure mechanisms of the NAD+-binding pocket, resulting in the identification of key interactions and allosteric communication pathways that initiate S1 and S2 openings and S1 and S2 closures. Importantly, these key residues are highly conserved among members of the Sir2 family, underscoring the significance of our work and suggesting that Sir2 proteins are governed by a universal deacetylation mechanism. Muvva et al. also identified some key residues in the CBL that govern NAM unbinding in SIRT1–3 through RAMD simulations, further highlighting the functional importance of the CBL. By addressing the atomic-level activity of the Sir2 loop region, our study provided new insights into broader roles of Sir2, including interactions with histones and mitochondrial DNA, with implications for interventions targeting cancer, aging, metabolic disorders, and other diseases.

However, it is crucial to acknowledge the limitations of our study, primarily its reliance on bacterial homologues of human proteins. Although the bacterial homologue of SIRT1 mimics interactions with human p53 and exhibits conserved key residues, nuanced differences between bacterial and human systems could have affected our findings, particularly regarding p53 interactions. Future simulations and experiments using human SIRT1 would offer a more relevant understanding of interactions between the loop and binding pocket, thereby enhancing the fidelity and applicability of our results to human biology and drug design. In addition, no available solved structures of p53/Sir2 or K^Ac382-p53/Sir2 complexes include the complete loop region, and our analysis therefore depends on computational modeling. High-resolution structures of these complexes would offer direct structural validation of the conformational states and communication pathways identified here. Although targeted mutational MD simulations would further strengthen the mechanistic insights, they fall outside the current scope. We nevertheless consider them a promising next step for future investigation. Moreover, previous studies have demonstrated that SIRT1 harbors a substrate-binding domain (SBD) implicated in substrate recognition, conformational stabilization, and allosteric regulation. By contrast, Sir2-af2 lacks this domain, underscoring a key structural divergence between the archaeal homologue and human SIRT1. Accordingly, further computational investigations specifically addressing the functional role of the SBD in SIRT1 are important to provide deeper mechanistic insight.

Our focus on specific regions within Sir2, particularly CBL opening/closure as characterized by S1/S1′ and S2, significantly advanced our understanding of Sir2 activation. The findings of this study lay the groundwork for future research to determine whether Sir2 function is controlled via a universal mechanism. Additionally, comprehensive testing of a wider range of substrates with Sir2, both computationally and experimentally, will be necessary to validate this mechanism. Comparative analyses of binding pocket areas in different Sir2 complexes and monitoring their conformational changes should further enhance our understanding of Sir2 activation and molecular interactions in living cells.

Finally, although our research focused on the p53/Sir2 complex, the saPaCS-MD technique employed to enhance sampling of opening/closure mechanisms holds promise for studies of other biological systems with similar mechanisms. By effectively sampling events between the open and closed states, saPaCS-MD offers a robust alternative to conventional MD simulations that could provide greater insights into complex biological processes.

MD input files, Python scripts and net entropy transfer files are deposited at https://github.com/Kitaolab/Sir2-p53.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.5c01755.

• Protocol for cMD simulations (S1); Free energy computation via MSM (S2); Entropy transfer calculation (S3); Occupancies of hydrogen bonds and salt bridges (Table S1); Average number of hydrogen bonds and salt bridges across different states (Table S2); Probability density distributions of S1, S1′, and S2 obtained from PaCS-MD simulations (Figure S1); Structural alignment of Sir2-Af2 and Sir2-Tm bound to acetylated p53 (Figure S2); Time evolution of surface areas S1/S1′/S2 (Figure S3); Projection of trajectories generated by saPaCS-MD simulations onto the principal components (Figure S4); Correlation analysis between surface areas and principal components (Figure S5); Implied time scale versus lag time plots for saPaCS-MD/MSM of apo Sir2, Sir2/KAc382-p53, and Sir2/p53 (Figure S6); Examples of implied time scale versus lag time plots for dPaCS-MD/MSM of Sir2/K^Ac382-p53 and Sir2/p53 (Figure S7); Free energy changes ΔW as a function of the center-of-mass distance d between Sir2 and K^Ac382-p53, and Sir2 and p53 (Figure S8); Macrostate assignment and transition time for Apo Sir2 and Sir2/p53 (Figure S9); Contact difference maps comparing Sir2 conformational states based on Cα distances (Figure 10); Multiple sequence alignment of Sir2 family members (Figure S11); S1′ and S2 Surface areas of Sir2 family members (Figure S12); Structures of Sir2/KAc382-p53/Carba-NAD84 and Sir2/p53/ OAADPr (Figure S13) (PDF)

• Movie of the S1 closure process from Open to Close1 (Movie S1) (MP4)

• Movie of the S2 closure process from Open to Close2 (Movie S2) (MP4)

• Movie of the S1 opening process from Close1 to Open (Movie S3) (MP4)

• Movie of the S2 closure process from Close2 to Open (Movie S4) (MP4)

Z.B., D.P.T., and A.K. planned this research and wrote the manuscript. Z.B. ran the simulations and performed the analysis. A.K. supervised this research.

This research was supported by MEXT/JSPS KAKENHI grants (nos. JP23H02445, JP23H04058, JP23H02424, JP24H02259, and JP24H01357 to A.K.; no. JP23K14154 to D.P.T.), JSPS Bilateral Program no. JPJSBP120236503 to A.K., and MEXT as “Program for Promoting Researches on the Supercomputer Fugaku” (Simulation- and AI-driven next-generation medicine and drug discovery based on “Fugaku”, JPMXP1020230120) to D.P.T.

The authors declare no competing financial interest.