Structures of archaeal DNA segregation machinery reveal bacterial and eukaryotic linkages

See allHide authors and affiliations

Science  04 Sep 2015:
Vol. 349, Issue 6252, pp. 1120-1124
DOI: 10.1126/science.aaa9046

Plasmid partitioning superstructure system

Partitioning and sharing DNA between dividing cells is critical for all domains of life. Prokaryotes must share certain plasmids as well as their genomic DNA to survive. Schumacher et al. studied the partition system that segregates a conjugative plasmid in the prokaryote Sulfolobus. The system consists of three proteins. AspA spreads along the plasmid DNA to create a protein-DNA superhelix. The ParA motor protein is linked to the protein-DNA superhelix through the ParB protein, which has structural similarities to eukaryotic centromere segregating proteins.

Science, this issue p. 1120


Although recent studies have provided a wealth of information about archaeal biology, nothing is known about the molecular basis of DNA segregation in these organisms. Here, we unveil the machinery and assembly mechanism of the archaeal Sulfolobus pNOB8 partition system. This system uses three proteins: ParA; an atypical ParB adaptor; and a centromere-binding component, AspA. AspA utilizes a spreading mechanism to create a DNA superhelix onto which ParB assembles. This supercomplex links to the ParA motor, which contains a bacteria-like Walker motif. The C domain of ParB harbors structural similarity to CenpA, which dictates eukaryotic segregation. Thus, this archaeal system combines bacteria-like and eukarya-like components, which suggests the possible conservation of DNA segregation principles across the three domains of life.

DNA segregation or partition is an essential biological process ensuring faithful genomic transmission. The best-understood segregation systems at the molecular level are those used by bacterial plasmids. These simplified systems consist of a DNA centromere; ParA nucleoside triphosphatase (NTPase); and centromere-binding protein (CBP), ParB (16). The most common partition apparatuses use Walker-box NTPases (7). The segregation systems that have been identified on bacterial chromosomes also harbor Walker NTPases, although their mechanisms are less clear (16). In contrast to bacterial partition, eukaryotic partition is highly complex. However, the linchpin in eukaryotic segregation is the histone protein, CenpA, which is deposited in place of histone H3 at DNA centromeres and dictates assembly of the segregation machinery (811). Although some progress has been made in understanding DNA segregation in eukarya and bacteria, virtually nothing is known about the molecular process of segregation in archaea, the third domain of life (1214).

To gain insight into the underpinnings of archaeal segregation, we performed a molecular dissection of the proteins encoded on the plasmid pNOB8 partition cassette harbored in Sulfolobus NOB8H2 (15). This cassette contains three open reading frames (ORFs): orf44, orf45, and orf46. orf46 and orf45 encode 315- and 470-residue proteins, respectively, which show 33 to 37% and 42 to 58% sequence similarity to bacterial ParA and ParB proteins. orf44 generates a 93-residue protein that shows no sequence similarity to any characterized partition protein. The organization of the orf44-parB-parA genes (fig. S1A) is reminiscent of bacterial partition cassettes; however, the cluster is tricistronic, unlike typical bicistronic bacterial systems (14). Hypothesizing that the pNOB8 centromere may be located either 5′ or 3′ of the cassette as in bacteria, we analyzed pNOB8 protein binding to these regions. Unlike its bacterial counterparts, pNOB8 ParB only bound DNA nonspecifically (fig. S1B). Orf44, however, bound the upstream DNA with high affinity [apparent dissociation constant (Kdapp) = ~50 nM] (Fig. 1A). Therefore, we named this CBP, archaeal segregation protein A (AspA). Deoxyribonuclease I (DNase I) footprinting showed that AspA interacted with a 23–base pair (bp) palindrome in the upstream region and that increasing AspA concentrations led to spreading around this site (Fig. 1B). Hence, like bacterial ParB proteins, AspA spreads nonspecifically to DNA adjacent to its centromere to form an extended “partition-complex.” Although poorly characterized, higher-order partition-complexes are central to segregation, as they mediate stabilizing, dynamic interactions with ParA assemblages (14). Indeed, the CBP-NTPase interaction is key to the partition process. Biochemical experiments, however, showed that AspA does not bind pNOB8 ParA. Rather, ParB bound both AspA [dissociation constant (Kd = 12 μM)] and ParA (Kd = 17 μM), which indicated that it acts as an adaptor (fig. S1C). Thus, the pNOB8 partition system is composed of the CBP AspA, adaptor ParB, and ParA NTPase. This finding prompted us to perform BLAST searches for the occurrence of aspA-parB-parA cassettes on archaeal genomes. The results revealed that this tripartite cluster of genes is widespread across different crenarchaeal genera and is harbored on both chromosomes and plasmids (fig. S2).

Fig. 1 Identification of pNOB8 centromere and AspA-centromere structure.

(A) AspA (0 to 1000 nM) was incubated with three DNA fragments upstream of its gene or the pNOB8 traG gene and subjected to electrophoretic mobility shift assays. White triangles denote location of the palindromic centromere. Arrows, unbound DNA; brackets, high-molecular-weight AspA-DNA complexes. The percentage of bound DNA was plotted against the concentration of AspA. (B) DNase I footprint identifying the DNA site bound by AspA (left, 0 to 750 nM) and its ability to spread at higher concentrations (right, 0 to 3000 nM). First lane, AG ladder; last lane, bovine serum albumin control. The inverted centromere repeat is in green and the aspA gene in blue. Below footprints, the sequence shaded in gray indicates the extent of the AspA footprint on top and bottom strands at low-to-medium protein concentrations. (C) (Top) AspA–32-oligomer (AspA-32mer) structure. Three interacting AspA dimers coat the DNA, which generates a continuous superstructure. (Bottom) Close-up showing contacts between recognition helices occupying the same major groove (31). (D) AspA N-terminal arm-DNA contacts. (E) Schematic diagram showing AspA-centromere interactions (31).

The AspA structure was obtained to gain insight into its function (table S1 and fig. S3A). The structure consists of a winged helix-turn-helix module followed by a C-terminal dimerization helix. Biochemical data supported the structural analysis that AspA is dimeric (fig. S3B). The AspA fold is distinct from bacterial CBPs and shows structural similarity with the PadR family of transcription regulators, found in bacteria and archaea (16). Three structures were determined of AspA bound to centromere DNA sites (fig. S3C and table S1). The structures all contain multiple interacting AspA dimers bound to DNA, that, when extended, lead to the generation of the same periodically arranged left-handed protein-DNA superhelix (Fig. 1, C to E, and Fig. 2A). This superhelix bears general resemblance to the left-handed Vibrio cholera ParA2–adenosine triphosphate–DNA (ParA2-ATP-DNA) filament (17). The protein and the DNA cannot be distinguished in the low-resolution ParA2-DNA electron microscopy structure, which makes a more direct comparison difficult. However, the AspA and ParA2 protein folds and superhelices are distinct; AspA and ParA2 are PadR and Walker-box proteins, respectively; and the ParA2-ATP-DNA superhelix has a pitch of 120 Å compared with 70 Å for the AspA-DNA superhelix. The high resolution of the AspA-DNA structure allows details of binding and superhelix formation to be discerned. Specifically, the insertion of two AspA α3 recognition helices of adjacently bound AspA dimers into each expanded major groove (~13 Å compared with 11 Å for B-DNA) leads to superhelix generation. The small Ala53 side chain appears key in the concurrent occupation of both helices in the same major groove. Leu52-Ala53 and Glu54-Lys59 contacts bolster this interaction (Fig. 1C). To be capable of DNA spreading, a protein must show adaptability in DNA binding. Almost all DNA base and phosphate contacts come from two adaptable AspA elements: the N-terminal arms and glutamines from the recognition helices (fig. S3D). The AspA eight-residue N-terminal arms fold only upon DNA binding and adopt a surprising range of conformations, which allows them to insert into either the major or minor grooves, where they make base interactions, phosphate contacts, or both (Fig. 1D). Consistent with the superhelical structure, isothermal titration calorimetry (ITC) experiments revealed a stoichiometry of wild-type AspA (wtAspA) dimer to 32-oligomer duplex of 3:1. By contrast, a AspA(A53K) mutation, which replaced the small Ala53 side chain with a bulky residue, bound DNA with a stoichiometry of 1:1 (Fig. 2B). Further, DNase I protection revealed that AspA(A53K) was unable to spread on DNA (Fig. 2C). Thus, multiple data support the AspA-DNA superhelix structure.

Fig. 2 AspA-DNA superhelix formation.

(A) AspA-DNA superstructure. The centrally bound dimer is magenta, and dimers spreading in the 5′ and 3′ directions are green and blue. (B) (Top) ITC of wtAspA binding to the 32-oligomer (32mer) DNA, revealing a binding stoichiometry of three AspA dimers to one DNA duplex. (Bottom) ITC analysis of AspA(A53K)-32mer interaction showing a stoichiometry of one AspA dimer to one DNA duplex. (C) DNase I footprints performed with wtAspA and AspA(A53K) binding to the pNOB8 par upstream region (as in Fig. 1B). Unlike wtAspA, AspA(A53K) does not spread.

The pNOB8 ParB adaptor was found to be composed of two domains, an N domain (ParB-N; residues 1–320) and a C domain (ParB-C; residues 370–470). Full-length ParB bound ParA in the presence or absence of ATP. However, neither ParB-C nor ParB-N bound ParA, which indicated that the flexible, ~50-residue interdomain linker mediates ParA binding (fig. S4, A to C). AspA did not bind ParB-C (fig. S5A) but bound ParB-N, and ITC showed the binding stoichiometry was 1:1, one AspA subunit to one ParB-N molecule (Fig. 3, A and B). The nonspecific DNA binding function of ParB was mapped to ParB-C (fig. S5B). Thus, pNOB8 ParB carries out multiple adaptor functions: binding AspA, ParA, and nonspecific DNA (fig. S4C). We next performed structural studies on ParB-N. As pNOB8 ParB-N crystals diffracted poorly, the 98:2 Sulfolobus solfataricus strain 98/2 chromosomal ParB (40% identity to pNOB8 ParB) N domain was used for crystallographic analyses (table S2). The ParB-N structure contains two subdomains separated by a central helix (Fig. 3C). Dali searches revealed weak structural similarity between ParB-N and one protein; the N-terminal domain of the bacerial Thermus thermophilus ParB homolog, Spo0J [root mean square deviation (rmsd) = 3.3 Å] (Fig. 3C, right) (18). However, there are large structural differences in the N-terminal subdomains and central regions of these proteins. The central region contains a loop in ParB-N and a helix-turn-helix (HTH) in Spo0J (Fig. 3C). The lack of an HTH in ParB-N is consistent with its inability to bind DNA and suggests that this domain must have evolved for an alternative function or functions, e.g., binding to AspA.

Fig. 3 Structure of ParB-N and assembly onto the AspA-DNA superhelix.

(A) Increasing concentrations of wtParB-N (blue) or ParB-N α11-α12-α13 truncation mutant (red) were added to a preformed AspA-DNA complex to analyze binding. mP, millipolarization. (B) ITC analyses revealing that ParB-N does not bind DNA (left) but binds AspA-DNA (right). (C) (Left) pNOB8 ParB-N structure. (Right) Overlay of ParB-N onto T. thermophilus Spo0J. Notable structural differences are denoted by asterisks. (D) AspA–ParB-N model obtained by docking the AspA and ParB-N crystal structures into the SAXS envelope. (E) AspA–DNA–ParB-N model obtained by fitting the ParB-N–AspA model onto the AspA-DNA superhelix. The AspA-DNA superhelix is blue, and ParB-N molecules are shown as coarse cartoon surfaces, with individual molecules colored red and magenta.

Small angle x-ray scattering (SAXS) (1922) was used to study the AspA–ParB-N interaction and revealed a symmetric ellipsoid-like structure for the complex that could be fit by placing an AspA dimer in the center flanked by ParB-N molecules [radius of gyration (Rg) of model = 34.2 Å compared with 33.2 ± 0.4 Å from experimental data] (supplementary Materials and Methods). In this model, which is consistent with the ITC stoichiometry of 1:1, the AspA C-terminal helices are grasped by ParB-N helices α11-α12-α13 (Fig. 3D), which dimerize in the model; the same α11-α12-α13 dimer was observed in ParB-N structures. Note that the AspA–ParB-N model also leaves the N-terminal face of the AspA dimer free to contact DNA, consistent with our data (Fig. 3D). The AspA dimer–ParB-N SAXS model can be docked onto each AspA dimer in the AspA-DNA superhelix, which leads to the generation of a multiprotein superstructure, in which ParB-N molecules conform to AspA-DNA superhelical grooves (Fig. 3E). Multiple modeling attempts, without taking into account the SAXS data, revealed that this organization provided not only the best fit to the AspA-DNA superstructure but also the only one without a clash. These data suggest that the specific AspA-DNA superhelical structure may function as a template for ParB-N binding (Fig. 3E, left). To test the AspA–DNA–ParB-N model, we generated a stable ParB-N truncation mutant (fig. S6), lacking helices α11-α12-α13, which the model indicates as important in AspA binding. ITC and fluorescence polarization (FP) experiments (Fig. 3A) revealed no binding by the mutant.

The AspA–DNA–ParB-N superstructure exposes the ParB C terminus at the surface, which would project the long linker and attached ParB-C into the solvent, which allows these regions to contact ParA and nonspecific DNA, respectively. The ParB-C structure was solved and consists of five helices (fig. S7). ParB-C forms a tight dimer (burying 2530 Å2), which was supported by biochemical analyses (Fig. 4A and fig. S7A). Notably, ParB-C shows structural similarity to histones, including CenpA (Fig. 4A and fig. S7B). This is important because CenpA is the histone variant that marks centromeres in eukaryotes (8). Although the ParB-C dimer is somewhat distinct from histone dimers, it contains an electropositive surface pattern similar to histones and as noted, it binds DNA (fig. S5B). A ParB-C–DNA model constructed by superimposition with nucleosome structures shows that the DNA is proximal to the ParB-C basic patches (Fig. 4B). A conserved feature of CBPs proteins that work in concert with Walker-box NTPases is their ability to both spread onto and bridge DNA (23, 24). The presence of many dynamic CBP-ParA attachments could maintain association of CBP-centromere complexes with ParA but simultaneously allow movement. Unlike bacterial CBPs, the pNOB8 CBP, AspA, forms a linear partition complex and shows no bridging capability. The nonspecific DNA binding function of ParB-C could fulfill that role.

Fig. 4 pNOB8 ParB-C contains a eukaryotic histone–CenpA-like fold, and ParA harbors a bacterial Walker-box fold.

(A) Superimposition of the ParB-C and Kluyveromyces lactis Cse4-CenpA structures (rmsd = 2.5 Å for Cα atoms). (B) ParB-C modeled onto DNA to show its electrostatic surface. (C) pNOB8 ParA–AMP-PNP subunit structure. (D) Superimposition of apo and AMP-PNP–bound ParA subunits showing relation of other subunit.

Partition mechanisms are dictated by their motor (NTPase) proteins, which bind CBP–centromere partition complexes. The pNOB8 ParA structure shows strong similarity to bacterial Walker-box ParA proteins (2527) (Fig. 4C, fig. S8, and table S2). Two conformations of pNOB8 ParA were obtained: apo and adenylyl-imidodiphosphate (AMP-PNP)–bound (fig. S9, A and B). Comparison of the AMP-PNP and apo ParA dimers reveals that the so-called signature lysines make the expected intersubunit phosphate contacts in the AMP-PNP–bound state, which stabilizes dimerization and enhances ATP hydrolysis. However, these residues are too distant in the apo conformation to make the same interaction (Fig. 4D). Thus, AMP-PNP (or ATP) binding to pNOB8 ParA appears to lock in a dimer state optimal for adenosine triphosphatase (ATPase) activity. Both ATP binding and hydrolysis are required for partition by Walker-box ParAs, and current data suggest that these proteins move dynamically, en masse, on the nucleoid DNA, which serves as a substratum (28, 29). Cycles of ParA movement are driven by CBP binding, which stimulates the ATPase activity of ParA. This represents a viable segregation mechanism for archaea, which like bacteria, lack a nuclear membrane. Consistent with this, pNOB8 ParA bound DNA nonspecifically in an ATP-dependent manner, similar to bacterial ParAs (figs. S9C and S10).

In summary, this work reveals important molecular insights into DNA segregation in archaea. This includes a delineation of the partition protein structures and higher-order complexes (fig. S10). The pNOB8 ParB component was shown to harbor structural similarity to histone-CenpA, which functions as the eukaryotic CBP. On the other hand, pNOB8 ParA harbors a Walker motif similar to that of bacterial ParAs. Recent data suggest that S. solfataricus itself uses a chromosomal partition system with a predicted Walker-box NTPase (30). Thus, these findings suggest that Walker-box cassettes may be the most ubiquitous DNA segregation system in nature and also suggest the possible conservation of general segregation principles across the three domains of life.

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Tables S1 and S2

References (3241)


  1. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.
  2. ACKNOWLEDGMENTS: This work was supported by NIH grant GM074815 (M.A.S) and UK Biotechnology and Biological Sciences Research Council (BB/F012004/1) and Leverhulme Trust (RPG-245) grants (D.B.). SAXS and crystallographic data were collected at the Advanced Light Source (ALS). ALS is a national user facility operated by Lawrence Berkeley National Laboratory on behalf of the Office of Basic Energy Sciences, U.S. Department of Energy (DOE), through the Integrated Diffraction Analysis Technologies program, supported by the DOE Office of Biological and Environmental Research. Additional support comes from MINOS at National Center for Biotechnology Information, NIH (R01-GM105404). We thank Q. She for the kind gift of the Sulfolobus NOB8H2 strain and N. Matamala for initial contributions to the project. Coordinates and structure factors have been deposited with the Protein Data Bank, accession numbers 4RS7, 4RS8, 5K5O, 5K5R, 5K5D,5K5Q, 5K5A, and 5K5Z.
View Abstract

Navigate This Article