key: cord-0749480-tzxnlw0j
authors: Appel, C. Denise; Feld, Geoffrey K.; Wallace, Bret D.; Williams, R. Scott
title: Structure of the sirtuin‐linked macrodomain SAV0325 from Staphylococcus aureus
date: 2016-07-07
journal: Protein Science
DOI: 10.1002/pro.2974
sha: 9b2d192f226d6d1d84adf5226fa562758ffe9055
doc_id: 749480
cord_uid: tzxnlw0j

Cells use the post‐translational modification ADP‐ribosylation to control a host of biological activities. In some pathogenic bacteria, an operon‐encoded mono‐ADP‐ribosylation cycle mediates response to host‐induced oxidative stress. In this system, reversible mono ADP‐ribosylation of a lipoylated target protein represses oxidative stress response. An NAD(+)‐dependent sirtuin catalyzes the single ADP‐ribose (ADPr) addition, while a linked macrodomain‐containing protein removes the ADPr. Here we report the crystal structure of the sitruin‐linked macrodomain protein from Staphylococcus aureus, SauMacro (also known as SAV0325) to 1.75‐Å resolution. The monomeric SauMacro bears a previously unidentified Zn(2+)‐binding site that putatively aids in substrate recognition and catalysis. An amino‐terminal three‐helix bundle motif unique to this class of macrodomain proteins provides a structural scaffold for the Zn(2+) site. Structural features of the enzyme further indicate a cleft proximal to the Zn(2+) binding site appears well suited for ADPr binding, while a deep hydrophobic channel in the protein core is suitable for binding the lipoate of the lipoylated protein target.

The reversible post-translational transfer of adenosine diphosphate ribose (ADPr) moieties to proteins enables cells to dynamically control response to external stimuli. In bacteria, ADP-ribosyltransferases (ARTs) are the primary family of proteins responsible for ADP-ribosylation of target proteins. 1, 2 Sirtuin proteins also have established functions in diverse cellular processes such as aging, metabolic regulation and gene silencing. 3 While they typically catalyze protein deacetylation reactions, members of the Sirtuin family including the macrodomain-linked sirtuins (SirTMs) have been shown to exhibit ADPribosylating activity. [3] [4] [5] Both ARTs and SirTMs utilize a bound nicotamide adenosine dinucleotide (NAD 1 ) cofactor, and in their reactions release nicotamide while covalently linking ADPr to a substrate. [1] [2] [3] [4] [5] Substrates may be either a specific amino acid (usually a glutamate/aspartate or lysine) on the target protein, or in the case of some eukaryotic ARTs, another ADPr though a glycosidic ribose-ribose bond forming linear and branched polymers. Structurally conserved in all domains of life, ARTs are an ancient protein class with diverse modifying abilities and target proteins. Bacterial ARTs act as toxins, modifying host cell signaling proteins, resulting in cell death and disease. [6] [7] [8] [9] [10] ARTs primarily transfer a single ADPr, a reaction referred to as mono-ADPribosylation (MARylation). In eukaryotes, ARTs have evolved into two distinct classes, those that resemble bacterial, MARylating ARTs and those capable of labeling target proteins with structurally and functionally diverse ADPr polymeric chains. 11, 12 The latter enzymes are often referred to as poly-ADPr polymerases (PARPs), and control diverse cellular processes including transcription and DNA repair through poly-ADPribosylation (PARlyation). [11] [12] [13] [14] The existence and coordination of enzymes capable of both mono ADP-ribosylation and poly ADPribosylation of target proteins paints a picture of a complex and regulated network of ARTs that can quickly and specifically alter cellular physiology post-translationally. 2 A number of enzymes complete the ADPr modifying cycle by catalyzing the removal of ADPr from poly-ADPr-targeted proteins. 15 Many of these proteins contain a macrodomain fold, which enables selective recognition of ADPr 16 and discrimination between mono-and poly-ADPr. 15 Initially discovered associated with peptidases in viruses 17 macrodomains have been found in all forms of life, but it was not until crystal structures of macrodomain proteins from the thermophile Archaeoglobus flulgidus 16, 18 and subsequently coronavirus 19 were their ADPr binding functions understood. Possessing a central six or seven stranded mixed b-sheet sandwiched by four to five a-helices, the macrodomain presents an L-shaped cleft that envelops the ADPr moiety largely through conserved protein backbone interactions, as well as aspartate-N6 adenosine and asparagine-O2 0 ribose interactions. Macrodomain proteins also function as O-acetyl-ADPr deacetylases, removing the acetyl moiety from the common reaction product of sirtuins. 20 The macrodomain-containing family includes the Poly-ADPr glycohydrolases (PARGs). [20] [21] [22] The catalytic functions of two additional human macrodomain proteins capable of reversing the glutamate-linked mono-ADPr, MacroD2, 23, 24 and TARG1 (Terminal ADP-Ribose Glycohydrolase 1), 25 have also been reported. Thus, the ADP-ribosylation cycle, including MARylation and/or PARylation, can be completed via macrodomain-mediated removal of the terminal ADPr.

Recently, an operon-encoded reversible MARylation system was identified in the pathogenic bacteria Staphylococcus aureus and Streptococcus pyogenes. 5 This system regulates the pathogen response to hostinitiated oxidative stress through lipoate protein ligase A (LplA2) transfer of a lipoate to a glycine cleavage system H-like (GcvH-L) target protein [ Fig.  1(A) ]. At the center of this regulation is a MARylation/ deMARylation cycle, where a novel class of macrodomain-linked sirtuins (SirTM) catalyzes the NAD 1 -dependent transfer of a mono-ADPr to GcvH-L, while a macrodomain protein removes the moiety [ Fig. 1(A) ]. Rack et al. showed that only a lipoylated GcvH-L could be MARylated by SirTM. Lipoylated GcvH-L is also efficiently deMARylated by the macrodomain (SauMacro for S. aureus and SpyMacro for S. pyogenes). Altogether, the study points to the modification of a shuttle protein (GcvH-L) with a redoxactive molecule (lipoate) under reversible control of a sirtuin/macrodomain-catalyzed negative feedback mechanism, where MARlyated GcvH-L represses and deMARlyated GcvH-L activates pathogen redox defense. 5 While the molecular basis of SirTM function has been established, 5 the structure and function of the SauMacro protein remains less well defined. Here, we report the crystal structure of SauMacro (a.k.a. Sav0325) to 1.75-Å resolution. Through sequence and structural analysis, we demonstrate that the SirTM-linked macrodomain protein family consists of a core macrodmain fold decorated by an Nterminal 3-helix bundle which scaffolds a novel Zn 21 coordination loop proximal to the ADPr binding site. SauMacro is further characterized by a conspicuous hydrophobic protein core cavity that is suitable for engagement of the lipoylated target substrate. Structures were aligned using MatchMaker in Chimera. 26 SauMacro Zn 21 is rendered as a sphere (magenta). All figures were prepared in Chimera 26 unless otherwise indicated.

Overall structure of the SAV0325 macrodomain Crystallization of recombinant SauMacro was facilitated by limited trypsin digestion of full-length Sau-Macro immediately prior to crystallization. For structure solution, an unpublished structure of the Eschericia coli macrodomain fold protein (RCSB entry 1SPV) was successfully used as a molecular replacement search model. The asymmetric unit contains a single SauMacro molecule with continuous electron density in two segments corresponding to residues 1 to 22 (plus an N-terminal non-native "GSHMAS" sequence from the expression tag) and residues 30 to 263 of SauMacro [ Fig. 1(B) , Table I ]. We hypothesize that trypsin cleavage occurred in the arginine rich disordered a1 0 -a2 0 linker region (amino acids [23] [24] [25] [26] [27] [28] [29] [30] , and that this aided in crystallization.

The SauMacro macrodomain structure is characterized by the canonical a/b macrodomain fold (six-stranded mixed beta sheet (b1-b6) that is sandwiched by five alpha helices a1-a5), and an added N-terminal helical domain. Overall, the macrodomain core of SauMacro resembles other macrodomain structures, and is most closely related to MacroD2 (RMSD of 1.6 Å ) [ Fig. 1 (C), Table II ]. Three important differences distinguish SauMacro from other previously determined macrodomain structures. First, SauMacro contains an aminoterminal extension comprised of an antiparallel 3helix bundle (a1 0 -a3 0 , residues 1-66), which is structurally distinct from the two amino-terminal helices found in MacroD2 and MacroD1 [ Fig. 1 (C), PDB: 2X47]. 20 Other macrodomain structures, including human TARG1 (PDB: 4J5S) 25 and AF1521, 16,18 also Table II] .

Notably, in the absence of 80 amino-terminal sequence, proteins with high sequence homology to SauMacro also lack the Zn 21 binding motif. Thus the three-helix bundle N-terminal extension appears to provide an extended structural scaffold for the Zn 21 -binding b2-a1 insertion loop. The proximity of the Zn 21 -binding motif to the predicted ADPr pocket implies that it may play roles in substrate binding and catalysis. The conserved residues C113, H118, and C120 mediate Zn 21 -coordination [ Fig. 2(A) ]. To validate the identity of the metal, we conducted X-ray fluorescence measurements on purified SAV0325 protein [ Fig. 2(C) ]. This analysis indicates that Zn is abundant in our protein purifications, consistent with Zn 21 modeled in the crystal structure. We also note that binding site geometry is consistent with a Zn 21 binding as confirmed by the "checkmymetal" (CMM) binding site validation server 27 (http://csgid.org/csgid/metal_sites/ ). The loop containing the 113-CxxxxHxC-120 Zn 21 binding motif appears to be an insertion relative to other known macrodomain structures, all of which replace this Zn 21 -binding sequence with a polyglycine linker. An inspection of the surface of Sau-Macro relative to other macrodomains containing ADPr reveals that the Zn 21 -binding motif does not occlude the canonical ADPr substrate-binding cleft. Interestingly, Asp55 of a symmetry related SauMacro molecule completes the tetrahedral coordination shell of the bound Zn 21 [ Fig. 2(B) ]. In the case of SpyGcvH-L, the MARlyated residue is expected to be either and Asp or Glu, 5 so the observed Asp55-Zn 21 interaction might reflect a possible substratebinding mode of the ADP-ribosylated target. Thus, we hypothesize that SauMacro and related sirtuinlinked macrodomains use this Zn 21 site to coordinate substrate binding and/or electron movements during catalysis.

ADPr was added to the crystallization solution, however no evidence for bound ligand was observed in our electron density maps. Notably, the loop between a4 and b6 is found in a conformation that partially occludes the predicted ADPr binding site, in a mode reminiscent of apo MacroD1. The CASTp server identifies a large contiguous solvent-accessible surface pocket in proximity to the Zn 21 site [Fig. 3(A-D) ]. This pocket comprises over 1400 Å 2 of surface area, and can be divided into two main sections connected via a narrow tunnel: one adjacent to the Zn 21 and the others containing the conserved acidic residue D93. Mapping the ADPr from the MacroD2 structure (PDB: 4IQY) onto this surface places the terminal ribose (ribose'') in the Zn 21 -adjacent pocket and the adenosine near D93. Electrostatic surface analysis implicates the surface around the Zn 21 pocket as electropositive, while the surface that would presumably contact the adenosine as electronegative [ Fig. 3(B) ]. Taken together, we infer that the identified pocket could accommodate the ADPr, consistent with SauMacro de-ADP-ribosylation function. 5 Next, we structurally aligned a number of similar macrodomains predicted by the DALI server to have similar structural features (Z score > 19) [ Table II , Fig. 3(C) ]. All of the compared structures contain an ADPr or ADPr-like bound substrate molecule in a similar spatial orientation on the protein surface. In considering the structurally similar macrodomains, many of the residues involved in substrate recognition are conserved in SauMacro [ Fig. 3(C,D) ]. Important electrostatic interactions include Asp93, which could hydrogen bond to adenine N6 amino group, as well as N107 and D122, both of which would contact the O2 0 hydroxyl of the terminal ribose. Substrate-bound macrodomains contain an aromatic residue that provides p-stacking interactions with the adenosine base; in SauMacro, F249 may perform this role [ Fig. 3(D) ]. C209, as well as the hydrophobic residues I94, I121, L187, and I211 line the pocket that could accommodate ADPr. G214 is a well-conserved residue, and the presence of a bulky side chain at this position would likely occlude ADPr phosphate binding. Some backbone rearrangements of SauMacro would be necessary to prevent clashes with ADPr, especially the loop connecting b5 and a4. Overall, the surface properties of this cleft, strong sequence homology to known ADPrmacrodomain complexes, and the previous report that SauMacro acts as an ADPr hydrolase 5 strongly implicate that ADPr binds in the predicted pocket.

In addition to binding and hydrolyzing ADPr, Sau-Macro, and other related macrodomains occupying LpA2 and SirTM-containing operons are expected to recognize lipoate. Efforts to co-crystallize SauMacro with lipoate were unsuccessful and our structure does not contain lipoate. However, we identify a conspicuous deep hydrophobic pocket that both spatially and electrostatically may accommodate the redox-active molecule [ Fig. 4(A) ]. This pocket is part of the surface identified by CASTp and resides in close proximity to the proposed ADPr binding site [ Fig. 4(B) ]. One face of the pocket contains Table II ). Sequence numbers and secondary structure assignments correspond to SauMacro. Red outlines indicate identical conservation, red letters indicate strong conservation (0.8 threashold). Yellow stars indicate Zn 21 binding residues in SauMacro. AteMfs1 and CalMfs1 contain additional carboxy-terminal sequence corresponding to a fused sirtuin protein. EclMacro and YmdB lack amino-terminal extensions and the conserved Zn 21 binding site.

residues 169-TVGPQ-173, which is highly conserved among macrodomains with high sequence homology to SauMacro [ Fig. 2(D) ]. The pocket is lined with highly conserved hydrophobic residues A105, I121, L187, F208, and I211, as well as Y191, which we predict would stack against the lipoate dithiolan ring. At the bottom of the pocket are two cysteines on the same face of helix a3, C190, and C194, which would lie within disulfide bonding distance to the two thiolates of the lipoate. Finally, the carboxyl face of the molecule, which would be covalently attached via an amide bond to a lysine of the target protein, extrudes out from the hydrophobic pocket, in potentially close contact to the conserved residue Q173.

The loop connecting b5 and a4 adopts three different conformations among the MacroD2, MacroD1, and SauMacro structures [ Fig. 4(B) ]. MacroD2 (PDB: 4IQY) and MacroD1 (PDB: 2X47), which are both ADPr-binding proteins, have very similar overall folds (RMSD of 0.85 Å over 210 Ca's), and this loop represents the most significant structural deviation between the two structures. Additionally, Mac-roD2 was determined in complex with ADPr, while MacroD1 is a ligand-free structure. Jankevicius et al. noted that an apparent 13 Å shift (5.3 Å for Ca positions) would be required to bring F272 in MacroD1 from an "open" conformation to the closed conformation occupied by Y190, the structurally homologous residue in MacroD2, stacks against the Poisson-Boltzman electrostatics for SauMacro are rendered as surfaces, ranging from electropositive (blue), neutral (white) to electronegative (red). C. Structure-specific sequence alignments of SauMacro with related macrodomains (see Table II ). Dark green outlines indicate identical homology, while light green outlines indicate highly similar homology (0.8 threashold). Structures aligned with DALI server. 29 D. Conserved residues (green) mapped onto SauMacro structure (gray). SauMacro rendered as ribbons with residues colored as in C. The position of ADPr from MacroD2 (yellow), as well as SauMacro residues predicted to interact with ADPr are rendered as sticks. Zn 21 (magenta) is shown as a sphere. Inset: Closeup view of ADPr binding site within the dashed box is presented. distal ribose of ADPr. 23 In SauMacro, the corresponding residue is likely F216, and its loop adopts an intermediate position, 10 Å (3.4 Å for Ca atoms) away from Y190 in MacroD2 [ Fig. 4(B) ]. SauMacro F216 is less likely to adopt the same closed conformation as Y190, as this position might occlude binding to the lipoate.

The close juxtaposition of the Zn-binding site, ADP-ribose pocket and the proposed hydrophobic lipoate-binding cavity suggests that the ADPr and lipoylation sites on the target GcvH-L proteins should be within close three-dimensional proximity of one another. The distance between the aspartic acid ligand bound in our structure (the proposed Asp-ADPr binding site) to the solvent accessible region of the lipoate binding site is 8 Å [Fig. 4(C) ]. Overall this geometry closely correlates with the spacing of the known GcvH-L ADPr modification (D27 of GcvH-L) and lipoylation (K56 of GcvH-L) sites [ Fig. 4(C) ].

The structure of SauMacro reported here reveals the first atomic details of a sirtuin-linked macrodomain. As a recently discovered class of oxidative-stress response regulators in pathogenic bacteria, the biochemical and biomedical relevance of these molecules are only beginning to be understood. Our structure suggests that SauMacro is a unique macrodomain protein with a novel Zn 21 binding insertion loop relative to other macrodomains. A threehelix bundle provides the structural scaffolding for this site, and sequence homology implies that this amino-terminal structure may be conserved for related macrodomains in bacteria and fungi. Through structural homology and analysis, we propose potential binding sites for ADPr and lipoate on SauMacro. Future co-crystal structures of SauMacro in complex with these substrates, or with the GcvH-L target protein, and in concert with biochemical analysis, will provide the foundation for potential novel antimicrobials targeting this stress response pathway.

N-terminally 6x His-tagged SauMacro (SAV0325) was expressed in from pET28a in E. coli Rosetta 2 (DE3) (Novagen). Terrific broth media was inoculated with saturated cell cultures and allowed to grow at 378C until OD 260 1.5, at which point the temperature was lowered to 158C, induced with 0.2 mM IPTG and further incubated overnight. Cells were pelleted, resuspended in 20 mM Tris pH 7.5, 150 mM NaCl, 10 mM Imidazole, sonicated and clarified by centrifugation. Affinity chromatography was accomplished by flowing clarified cell lysate over Ni-NTA resin. Bound protein was eluted by addition of imidazole and further purified by size-exclusion chromatography using a Superdex 75 column (GE Healthcare) equilibrated in 20 mM Tris pH 7.5, 500 mM NaCl, and 0.1% bME. Prior to crystallization a purified SAV0325 was subjected to limited trypsin digest at room temperature for 1 hr at a ratio of 650:1 (w/w) SauMacro to trypsin. Trypsinized protein was resolved by additional sizeexclusion purification, followed by a final cationexchange purification eluted in 20 mM Tris pH 7.5, 175 mM NaCl, and 0.1% bME. The resulting fractions from the single peak from the ion-exchange purification were pooled and concentrated to 25 mg/mL for crystallographic screening.

Crystals were grown at 48C using the sitting-drop vapor diffusion method, by mixing SauMacro (25 mg mL 21 protein plus 1 mM ADP-ribose) with 0.1 M CHES pH 9.5, 15% (v/v) ethanol in a 1:1 ratio (250 nL: 250 nL) using a Mosquito robot (TTP Labtech, Hertfordshire, UK). The resulting rod shaped crystals were cryoprotected in a solution of mother liquor supplemented with 26% (v/v) ethylene glycol and flash frozen in liquid nitrogen for subsequent data collection. Diffraction data were collected to 1.75 Å on a MAR 225 CCD detector at Southeast Regional Collaborative Access team (SER-CAT) 22-BM beamline at the Advanced Photon Source, Argonne National Laboratory. Data were indexed, integrated, and scaled using HKL2000. 30 The data were processed in P2 1 2 1 2 1 space group, with unit cell dimensions of a 5 46.823, b 5 47.853, c 5 134.950, a 5 b 5 c 5 908 (also see Table I ). X-ray fluorescence scans [ Fig. 2(C) ] on ammonium sulfate precipitated SAV0325 were performed and analyzed at SER-CAT 22-BM.

Crystallographic phases were determined by molecular replacement using 1SPV as a search model in Phaser. 31 Subsequent rounds of automated model building in PHENIX and manual model building in Coot 32 with refinement in PHENIX yielded the final model. Water molecules were placed automatically using PHENIX and checked manually in Coot. Significant positive density was found in between residues C113, H118, and C120, indicative of a metal ion coordination site. Based on ionic distances and coordination geometry, the metal was assigned as Zn 21 . Although the protein crystallization solution was supplemented with ADPr, no definitive difference electron density corresponding to this ligand was observed. Model validation was carried out automatically with Molprobity. 33 Data collection and refinement statistics are detailed in Table I . PyMOL was used for structuresequence conservation figures, while all other molecular graphics were produced with UCSF Chimera. 26 Solvent accessible surface pockets were calculated using the CASTp server. 28 To visualize electrostatic surfaces, the SauMacro coordinates were prepared for Poisson-Boltzman calculations using PDB2PQR, 34 and the calculations were performed with APBS. 35 

Homologous sequences were manually selected from a DELTA-BLAST search of the SauMacro protein sequence. These sequences were aligned using the MultAlign server 36 and visualized with ESPript. 37 Proteins with similar structural homology to SauMacro were identified with the Dali server, 29 and subsequent sequence alignments and figures were prepared with ENDscript. 37 All sequences used in this study are listed in Table II .

Coordinates and structure factors have been deposited in the Protein Data Bank as PDB entry 5KIV.

Toward a unified nomenclature for mammalian ADP-ribosyltransferases

The recognition and removal of cellular poly(ADP-ribose) signals

Sirtuins, metabolism, and DNA repair

Acetylationdependent ADP-ribosylation by Trypanosoma brucei Sir2

Identification of a class of protein ADPribosylating sirtuins in microbial pathogens

Diphtheria toxin-dependent adenosine diphosphate ribosylation of aminoacyl transferase II and inhibition of protein synthesis

Structure and activity of diphtheria toxin. I. Thiol-dependent dissociation of a fraction of toxin into enzymically active and inactive fragments

Evolution and mechanism from structures of an ADP-ribosylating toxin and NAD complex

A family of killer toxins. Exploring the mechanism of ADPribosylating toxins

Novel bacterial ADP-ribosylating toxins: structure and function

New insights into the molecular and cellular functions of poly(ADP-ribose) and PARPs

Structures and mechanisms of enzymes employed in the synthesis and degradation of PARP-dependent protein ADP-ribosylation

Structural basis for DNA damage-dependent poly(ADPribosyl)ation by human PARP-1

The rise and fall of poly(ADP-ribose): an enzymatic perspective

Recognition of mono-ADP-ribosylated ARTD10 substrates by ARTD8 macrodomains

The macro domain is an ADP-ribose binding module

Putative papain-related thiol proteases of positive-strand RNA viruses. Identification of rubi-and aphthovirus proteases and delineation of a novel conserved domain associated with proteases of rubi-, alpha-and coronaviruses

The crystal structure of AF1521 a protein from Archaeoglobus fulgidus with homology to the nonhistone domain of macroH2A

Crystal structures of two coronavirus ADP-ribose-1''-monophosphatases and their complexes with ADP-Ribose: a systematic structural analysis of the viral ADRP domain

Identification of macrodomain proteins as novel O-acetyl-ADP-ribose deacetylases

Substrate-assisted catalysis by PARP10 limits its activity to mono-ADP-ribosylation

Structure of mammalian poly(ADP-ribose) glycohydrolase reveals a flexible tyrosine clasp as a substrate-binding element

A family of macrodomain proteins reverses cellular mono-ADP-ribosylation

Macrodomain-containing proteins are new mono-ADP-ribosylhydrolases

Deficiency of terminal ADPribose protein glycohydrolase TARG1/C6orf130 in neurodegenerative disease

UCSF Chimera-a visualization system for exploratory research and analysis

Validation of metal-binding sites in macromolecular structures with the CheckMyMetal web server

CASTp: computed atlas of surface topography of proteins with structural and topographical mapping of functionally annotated residues

Dali server: conservation mapping in 3D

Processing of X-ray diffraction data collected in oscillation mode

PHENIX: a comprehensive python-based system for macromolecular structure solution

Features and development of Coot

MolProbity: all-atom structure validation for macromolecular crystallography

PDB2PQR: an automated pipeline for the setup of Poisson-Boltzmann electrostatics calculations

Electrostatics of nanosystems: application to microtubules and the ribosome

Multiple sequence alignment with hierarchical clustering

Deciphering key features in protein structures with the new ENDscript server