Crystal structures of HpSoj-DNA complexes and the nucleoid-adaptor complex formation in chromosome segregation.

ParABS, an important DNA partitioning process in chromosome segregation, includes ParA (an ATPase), ParB (a parS binding protein) and parS (a centromere-like DNA). The homologous proteins of ParA and ParB in Helicobacter pylori are HpSoj and HpSpo0J, respectively. We analyzed the ATPase activity of HpSoj and found that it is enhanced by both DNA and HpSpo0J. Crystal structures of HpSoj and its DNA complexes revealed a typical ATPase fold and that it is dimeric. DNA binding by HpSoj is promoted by ATP. The HpSoj-ATP-DNA complex non-specifically binds DNA through a continuous basic binding patch formed by lysine residues, with a single DNA-binding site. This complex exhibits a DNA-binding adept state with an active ATP-bound conformation, whereas the HpSoj-ADP-DNA complex may represent a transient DNA-bound state. Based on structural comparisons, HpSoj exhibits a similar DNA binding surface to the bacterial ParA superfamily, but the archaeal ParA superfamily exhibits distinct non-specific DNA-binding via two DNA-binding sites. We detected the HpSpo0J-HpSoj-DNA complex by electron microscopy and show that this nucleoid-adaptor complex (NAC) is formed through HpSoj and HpSpo0J interaction and parS DNA binding. NAC formation is promoted by HpSoj participation and specific parS DNA facilitation.

INTRODUCTION

Precisely replicated DNA segregation is necessary for accurate inheritance of genetic material (1). In bacteria, the partitioning (par) system (ParABS) is a highly-conserved process for the delivery and localization of chromosomes or low-copy-number plasmids in cells (2). The ParABS system comprises three elements: the NTPase motor protein (ParA: partitioning protein A); the centromere-binding protein (ParB: partitioning protein B); and a centromere-like DNA site (parS) (3,4). The general partitioning process involves specific binding of a parS site by ParB and assembly into a higher-order partitioning complex (2,4,5). The ParA dimer uses NTP-binding to trigger NTP hydrolysis, which drives segregation of newly-replicated ParB-parS-containing plasmids (6–8).

There are three classes of ParA NTPase motor protein: type I, a ‘deviant’ Walker-type ATPase; type II, an actin-like ATPase; and type III, a tubulin-like GTPase (8). Type I ParA can be further differentiated into types Ia and Ib depending on its structure and size (8). Type Ia ParA, such as Escherichia coli P1 ParA, has a non-specific DNA (nsDNA)-binding domain and a specific DNA-binding domain (9). Type Ib ParA—such as Streptococcus pyogenes pSM19035 Delta (δ, SpParA), Salmonella newport TP228 ParF (TP228 ParA), and Helicobacter pylori Soj (HpSoj) solely possess an nsDNA-binding domain (10–13). ParB comprises an N-terminal protein-protein interaction domain, a central helix-turn-helix DNA-binding domain, and a C-terminal self-dimerization domain (14).

Plasmid DNA stability is maintained by the ParABS system (15). In bacterial chromosome-encoded ParABS, Soj (sporulation protein J) and Spo0J (stage 0 sporulation protein J) are the homologs of the ParA and ParB proteins, respectively. In Thermus thermophiles and Bacillus subtilis, the ATPase activity of Soj is elevated by Spo0J upon addition of DNA (7,13,16). Soj exists as a monomer in the absence of ATP, but is dimeric when ATP is present. ATP-bound Soj dimer exhibits nsDNA-binding ability (7,16). Several studies have shown that DNA binding by ParA is essential for DNA segregation (17,18).

In the ParMRC segregation system of R1 plasmid, it was found that parC manages plasmid molecule pairing, which requires binding of ParR to parC, and the ParR-parC pair complex is enhanced by the ParM-ATP complex (19). The DNA partitioning system of pB171 plasmid involves ParA and ParB with two parC sites. It has been suggested that ParB mediates the pairing complex and the higher-order complex comprising several DNA molecules, and the N-terminus of ParB was observed to be essential for ParB-mediated pair formation (20). In the E. coli P1 ParABS system, ParB binds to parS DNA to form the pairing complex (the ParB-parS complex) before the nucleoprotein ParA–ParB–DNA complex is assembled (15,21–23). ATP regulates these interactions and the parS DNA stabilizes the nucleoprotein complex (22). The dynamic interaction between plasmid-bound ParB and nucleoid-bound ParA of the ParA–ParB–DNA complex is defined as the nucleoid-adaptor complex (NAC), which mediates cooperation between the plasmid and the nucleoid (22).

Two distinct models have been proposed for the ParABS system: (i) a filament model whereby the ParB partitioning complex (i.e. cargo, the ParB-parS DNA complex) is translocated by ParA, which is polymerized along the nucleoid to form a nucleoprotein filament (24,25); and (ii) a diffusion-ratchet model whereby the ParB-DNA cargo is associated with ParA via a concentration gradient of ParA-ATP in the nucleoid surface (26–29). Two diffusion-ratchet mechanisms have been proposed: (a) a DNA-relay mechanism involving elastic dynamics of the chromosome involved in the translocation of the ParB partitioning cargo (30,31); and (b) a hitch-hiking mechanism whereby ParA assembles in high-density DNA regions (HDRs) to guide ParB partitioning cargo delivery (32). The diffusion-ratchet model suggests that ParA-ATP needs to undergo a conformational change into a DNA-binding competent state, ParA*2-ATP2, for binding to the nucleoid (26), resulting in ParA*2-ATP2 attachment to the nucleoid (or localization at HDRs) to form a nucleoid matrix in which the association between plasmid-bound ParB and nucleoid-bound ParA is facilitated (26,27,29,32).

Two Walker-box DNA segregation systems were identified recently in archaeal plasmids and chromosomes, i.e., SegAB and AspA-ParBA machineries (33–36). The SegAB system was characterized in the thermophilic crenarchaeon S. solfataricus and consists of SegA and SegB proteins and a centromere-like region (34). SegA is an ortholog of the bacterial Walker-type ATPase ParA. Similar to the polymerization behavior of bacterial ParA, SegA assembles into polymers in the ATP-bound state. Archaea SegB is a site-specific DNA-binding protein that interacts with SegA and affects SegA polymerization (34). The AspA-ParBA machinery is encoded by the plasmid pNOB8 hosted by Sulfolobus NOB8H2 strain (35). It comprises AspA (archaeal segregation protein A, a centromere-binding component), ParA (NTPase), and ParB (an atypical ParB) (35). Archaea AspA is a site-specific DNA-binding protein with no sequence homology to any characterized segregation protein (35). The pNOB8 ParA possesses nsDNA-binding ability (37), and a multifaceted DNA-binding mode has been identified from the structure of pNOB8 ParA–AMPPNP–DNA (37). The pNOB8 ParB exhibits nsDNA-binding activity and functions as an adaptor protein to interact with AspA, ParA and nsDNA (35).

The crystal structure of the HpSpo0J-parS DNA complex of the ParABS system of H. pylori was previously determined (2), revealing that HpSpo0J molecular spreading and DNA bridging is mediated by the HpSpo0J N-terminus. Here, we report the crystal structures of HpSoj and its DNA complexes. HpSoj non-specifically binds to DNA through a basic binding patch comprising Lys199, Lys227, Lys230 and Lys247, and directly binds within the DNA minor groove at Asn196, Lys199 and Val229. Furthermore, we used electron microscopy (EM) to monitor nucleoprotein formation among HpSpo0J, HpSoj and DNA, allowing us to propose a potential NAC complex for the partitioning system in H. pylori.

MATERIALS AND METHODS

Protein expression and purification

The PCR product of the Hp1139 (HpSoj) gene (795 bp) from H. pylori 26695 genomic DNA was ligated into the BamHI and SalI restriction sites of expression vector pQE30 (Qiagen) that possesses a hexa-histidine tag at the N-terminus (13). The plasmid was then transformed into the expression host, E. coli strain SG13009. Recombinant HpSoj was grown in Luria-Bertani medium and induced overnight at 20°C by adding 1 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG). The Ni-NTA system (GE Healthcare) was used to purify HpSoj with elution buffer (20 mM Tris pH 8.0, 500 mM NaCl, 110 mM imidazole, 5 mM MgCl2 and 10% glycerol). The eluted proteins were applied to a Superose™ 12 gel filtration column (GE Healthcare) pre-equilibrated with the elution buffer and run at 0.5 ml·min−1. Molecular weight and purity of proteins were assessed by SDS-PAGE. HpSoj mutations (D41A, K199E, K230E, K199/230E and K199/227/230/247E) were generated by site-directed mutagenesis method and verified by DNA sequencing. Protein purification of these HpSoj mutants were similar to that of wild-type HpSoj.

PCR products of HpSpo0J and HpSpo0JN (residues 1–50) from H. pylori 26695 genomic DNA were ligated into the NcoI and XhoI restriction sites of the pET-28a(+) expression vector (Novagene) that contains a C-terminal hexa-histidine tag. The resulting two plasmids, pET-28a(+)-HpSpo0J and pET-28a(+)-HpSpo0JN, were transformed into E. coli strain BL21(DE3). The expression systems were incubated overnight at 37°C in Luria-Bertani medium. Cultures were induced by adding 1 mM IPTG and incubating at 37°C for three hours. Ni-NTA columns (GE Healthcare) were used to purify the two proteins. The eluted HpSpo0J and HpSpo0JN proteins were further purified by a Superose™ 12 gel filtration column (GE Healthcare).

DNA preparation

Double-stranded 24 and 55 bp parS DNA fragments were prepared using an equal molar ratio of two complementary oligonucleotides (the 16 bp parS site is underlined): 24 bp DNA fragment (5′-AGGGTGTTCCACGTGAAACAGGGA-3′) (for parS24), parS-contained 55 bp DNA fragment (5′-CTGAATCAGCAGTTGAATCAGAATGTTCCACGTGAAACAAAGAAAAAAGAACCTG-3′) (for parS55) and 55 bp DNA fragment (5′-CTGAATCAGCAGTTGAATCAGAATGTTC CACGTGAAACAAAGAAAAAAGAACCTG-3′) (for 55DNA). The double-stranded DNA (dsDNA) was annealed by heating to 95°C for 30 min and then cooled slowly to room temperature, before being stored at 4°C until use.

ATPase activity assay

The ATPase activity of HpSoj and HpSojD41A was determined using the malachite green method (38). The assay was performed with 4 μM (0.4 nmol) HpSoj or HpSojD41A and 4 mM (400 nmol) ATP in 100 μl buffer (20 mM Tris pH 8.0, 175 mM NaCl, 5 mM MgCl2 and 10% glycerol) at 37°C for two hours. We used 4 μM (0.4 nmol) HpSpo0J or HpSpo0JN in the ATPase assays, with (0.4 nmol) or without either parS55 or 55DNA. At various time-points (5, 20, 40, 60, 80, 100 and 120 min), 100 μl of the reaction solution was removed and immediately mixed with 900 μl acidic coloring solution (125 mM H2SO4, 0.5 mM ammonium-molybdate, 0.01 mM ascorbic acid, 0.04 mM potassium antimony III oxide tartrate). The color was allowed to stabilize for 30 min before absorbances of the antimony-phosphomolybdate complex were detected at 850 nm. Absorbance at 850 nm resulted from non-enzymatic hydrolysis of ATP was subtracted from that of each reaction mixture. We created a standard curve using a serial dilution of potassium phosphate (0–400 μM). A850 values were converted into moles Pi produced per mole of protein using phosphate absorbance as a standard. The control experiments in the presence of DNA, HpSpo0J and HpSpo0JN alone and in combination but without HpSoj were performed. The ATPase activities of these control experiments were near zero or undetectable.

Electrophoretic mobility shift assay

The DNA-binding ability of HpSoj was detected using EMSA. Reactions were performed in a volume of 24 μl in reaction buffer (20 mM Tris pH 8.0, 175 mM NaCl, 5 mM MgCl2 and 10% glycerol). Each reaction contained HpSoj (0–1000 pmol) and parS24 (24 bp) or parS55 (55 bp) (25 pmole) in various molar ratios of protein to DNA (0, 2, 10, 20 or 40), with or without 1 mM ADP or ATP. Reactions were incubated at 20°C for 20 min. Samples were analyzed on 4–15% gradient polyacrylamide gels (GoPAGE™ TGN Precast Gel, SMBiO) in 1× TGN buffer at 4°C for 150 min and visualized by GelRed™ Nucleic Acid Gel-staining (Biotium).

Microscale thermophoresis

HpSpo0JN proteins were labeled with the fluorescent amine-reactive red dye NT-647-NHS (NanoTemper Technologies) and then diluted to optimal fluorescence intensity (around 10,000 counts). Unlabeled HpSoj (214 μM) was serially diluted 1:1 in protein buffer (20 mM MOPS pH 7.0, 5 mM MgCl2, 110 mM imidazole, 250 mM NaCl, 50 mM arginine, 50 mM glutamic acid, 1 mM ATP) with 0.05% Tween20 over 16 tubes, each containing 10 μl protein. Aliquots were mixed with 10 μl of labeled HpSpo0JN (33 nM) before being transferred into hydrophilic glass capillaries. Measurements were performed at 25°C using 20% LED power and 60% MST power. In the negative control experiment, unlabeled lysozyme (250 μM) was also serially diluted and mixed with labeled HpSpo0JN. The instrument used was a NanoTemper monolith NT.115Pico and the data were analyzed from triplicate experiments.

Crystallization

HpSoj-ATP crystals were grown using HpSoj (8 mg/ml) with an extra 10 mM ATP as an additive. The reservoir solution contained Bicine pH 9.0, 8% Tacsimate, and 15% PEG3350. Crystals were obtained after one-week incubation at 10°C. HpSoj–ADP–DNA crystals were grown at a 2:1 molar ratio of HpSoj and parS24 (5′-AGGGTGTTCCACGTGAAACA GGGA-3′, underlined is the parS site). The reservoir solution contained 50 mM MES pH 5.6, 240 mM KCl, 10 mM MgCl2, and 6% PEG8000. HpSoj–ADP–DNA crystals were grown at room temperature over 3 days. HpSoj–ATP–DNA crystals were grown at a 5:1 molar ratio of HpSoj and parS24, and the reservoir solution contained 50 mM MES pH 5.6, 200 mM KCl, 10 mM MgCl2 and 4% PEG8000. Crystals of HpSoj–ATP–DNA were obtained within one day following incubation at 20°C.

Fluorescence polarization binding isotherms

The equilibrium DNA binding assays of HpSoj was done by fluorescence polarization binding isotherms. parS24 DNA substrates were fluorescently labeled on the 5′ end of DNA, which are enabled to measure the increase in fluorescence polarization of the HpSoj:DNA complex relative to the unbound DNA due to HpSoj binding. Serial dilutions of HpSoj proteins were made in 20 mM Tris-HCl pH 8.0, 175 mM NaCl, 5 mM MgCl2, 10% glycerol and 1 mM ATP before being incubated with 5 nM Cyanine-3 (Cy3)-labeled DNA at room temperature. DNA binding by HpSoj was determined by measuring the changes in fluorescence polarization using a Paradigm plate reader (Molecular Devices). Fluorescence polarization (FP) signal was read at 595 nm at an excitation of 535 nm and calculated by determining the concentration of HpSoj required to bind 50% of the Cy3-labeled DNA. The unbound state is represented by the fluorescence anisotropy of the Cy3-labeled DNA in the presence of buffer alone. The experiments of HpSoj mutant proteins (K199E, K230E, K199/230E, K199/227/230/247E) were performed as that of wild-type HpSoj. The average of three independent experiments is shown, with error bars representing standard deviations.

Data collection and structure determination

X-ray diffraction data for the HpSoj-ATP, HpSoj–ATP–DNA and HpSoj–ADP–DNA crystals were collected in the Super Photon ring-8 GeV (SPring-8), Japan, and the National Synchrotron Radiation Research Center (NSRRC), Taiwan. All datasets were processed by the HKL-2000 software (39).

The structural phases of HpSoj-ATP were determined by molecular replacement (MR) with Phaser (40), using TtSoj (PDB ID: 1WCV) (7) as a search model. Structural phases of HpSoj–ATP–DNA and HpSoj–ADP–DNA were solved by MR using HpSoj as the search model. Structural refinements were performed in Phenix (41), and structural model adjustment was carried out in COOT (42). X-ray diffraction data and structural refinements are summarized in Table 1.

Table 1.

X-ray diffraction data and refinement statistics of HpSoj and HpSoj–DNA complexes

Crystal	HpSoj-ATP	HpSoj–ADP–DNA	HpSoj–ATP–DNA
Data collection statistics
Source	Spring-8 44XU	NSRRC-TPS05A	NSRRC-TLS15A
Wavelength (Å)	0.90000	1.00000	1.00000
Space group	P212121	P1	P1
Resolution (Å)	1.9	2.5	3.4
Unit cell parameters
a (Å)	48.1	74.8	74.4
b (Å)	93.3	74.7	74.6
c (Å)	110.9	80.7	80.4
α (°)	90.0	71.5	71.4
β (°)	90.0	71.6	71.6
γ (°)	90.0	67.7	67.8
Number of reflections	181 905	97 062	32 909
Number of unique reflections	37 426 (3776)	49 136 (5041)	18 173 (2049)
Redundancy of reflection	4.9	2.0	2.0
Completeness (%), overall	93.6 (96.3)	96.0 (97.9)	88.0 (98.6)
I/σ(I), overall	23.7 (3.9)	13.4 (1.5)	8.5 (2.0)
R merge b (%), overall	7.3 (37.2)	5.0 (37.4)	13.3 (41.9)
Refinement statistics
Resolution (Å)	30.0–1.9	29.8–2.5	26.8–3.4
R-factorc/Rfreed (%)	16.6/20.2	20.4/24.7	20.1/25.0
Number of reflections used	36,400	49,125	17,784
Number of residues	528	1,056	1,056
Number of atoms
Protein	4106	8208	8208
DNA	-	984	984
ATP	62	108 (ADP)	124
Mg	2	4	4
Water	494	19	0
B-factor (Å2)
Protein	21	59	85
DNA	-	98	120
ATP	14	41 (ADP)	73
Mg	11	49	70
Water	29	-	-
RMSD bond lengths (Å)	0.008	0.017	0.012
RMSD bond angles (°)	1.267	1.928	1.698
PDB IDe	6IUB	6IUD	6IUC

aValues in parentheses are for the highest-resolution shell.

b R merge = Σ|I− |/ΣI, where I is the observed intensity and is the average intensity from multiple observations of symmetry-related reflections.

c R = Σ|Fobs− Fcalc|/ΣFobs, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively.

d R free was calculated with 5% of the total number of reflections randomly omitted from the refinement.

eProtein data bank identifiers for co-ordinates.

X-ray diffraction data and refinement statistics of HpSoj and HpSoj–DNA complexes

aValues in parentheses are for the highest-resolution shell.

b R merge = Σ|I− |/ΣI, where I is the observed intensity and is the average intensity from multiple observations of symmetry-related reflections.

c R = Σ|Fobs− Fcalc|/ΣFobs, where Fobs and Fcalc are the observed and calculated structure factor amplitudes, respectively.

d R free was calculated with 5% of the total number of reflections randomly omitted from the refinement.

eProtein data bank identifiers for co-ordinates.

Electron microscopy

For electron microscopy, parS550 (550 bp DNA with one parS site in middle), 180DNA (180 bp DNA) and 600DNA (600 bp DNA) were used. The parS550, 180DNA, and 600DNA fragments were amplified by PCR. The DNA-only control (Supplementary Figure S5) was examined by a Cytochrome C spreading technique (43). For the HpSpo0J:parS550 complex, we used 20 nM HpSpo0J and 2 nM parS550 in binding buffer (20 mM Tris pH 8.0, 150 mM NaCl, 5 mM MgCl2, 33 mM imidazole, 10% glycerol). For the HpSoj:180DNA complex, we used 80 nM HpSoj and 4 nM 180DNA in the presence of ATP (1 mM). HpSpo0J and parS550 DNA (molar ratio 10:1), HpSpo0J and 600DNA (molar ratio 10:1), or HpSoj and 180DNA (molar ratio 20:1) were incubated for 10 min at 37°C. To induce NAC formation, the HpSpo0J–parS550 complex and the HpSoj-180DNA complex were mixed together and incubated for 6 h at 37°C. The resulting protein-DNA mixture was placed on a clean Parafilm surface and was picked up on a carbon-coated grid before being negatively stained with 1% uranyl acetate. After the grid had been air-dried for one day, rotary shadowing was performed with the platinum-palladium source at an angle of 10 degree and a sample-to-source distance of 8 cm. The grid was coated with carbon and then visualized by electron microscopy. Images were captured by a Tecnai G2 Spirit TWIN (FEI Company) electron microscope at a magnification of ×15 000 at 80 kV. EM image quantifications were conducted using ImageJ. Image frames were randomly selected from different grids. The parS550, 180DNA, and 600DNA fragments were distinguishable by length. Numbers of various complexes in images are listed in Table 2 and Supplementary Table S3.

Table 2.

The counts and number of HpSpo0J–DNA, HpSoj–DNA and HpSpo0J–HpSoj–DNA

	Protein:parS550 complex	Protein:parS550 paired complex	Protein:parS550:180DNA complex	Higher-order Complex	Molecules Counted, % (n)
[HpSpo0J:parS550]	70.9% (144)	12.8% (26)	-	16.3% (33)	100% (203)
[HpSpo0J:parS550]+[HpSpo0J:180DNA]	85.3% (174)	6.4% (13)	1.5% (3)	6.9% (14)	100% (204)
[HpSoj:parS550]+[HpSoj:180DNA]	90.9% (140)	5.8% (9)	3.2% (5)	0.0% (0)	100% (154)
[HpSpo0J:parS550]+[HpSoj:180DNA]	72.3% (125)	6.9% (12)	18.5% (32)*	2.3% (4)	100% (173)

*Nucleoid–adaptor complex (NAC) formation.

The counts and number of HpSpo0J–DNA, HpSoj–DNA and HpSpo0J–HpSoj–DNA

*Nucleoid–adaptor complex (NAC) formation.

RESULTS

Characterization of HpSoj

HpSoj belongs to the ParA superfamily. We present a multiple sequence alignment in Supplementary Figure S1 of the ParA superfamily from the bacterial par system (7,9,11,16,25) and of archaeal Sulfolobus NOB8H2 pNOB8 ParA (35). HpSoj shares a sequence identity of 22–48% with the ParA superfamily. In addition, proteins of the ‘deviant’ Walker A motif family with different biological functions (44–47) are also shown in Supplementary Figure S1. HpSoj shares a sequence identity of 18–28% with the ‘deviant’ Walker A motif superfamily.

We assayed the ATPase activity of HpSoj by the malachite green method (37) (Figure 1A). HpSoj achieved ATP hydrolysis of 1.0±0.3 moles Pi released/mole HpSoj·hour−1. The activities show slightly increase of 1.4- and 1.3-fold with specific DNA (parS55) and non-specific DNA (55DNA), respectively. However, the ATPase activity of HpSoj with HpSpo0J reveals an increase of 3.8-fold using parS55 but only 1.2-fold with 55DNA. The ATPase activity of HpSoj was stimulated when HpSpo0J binds to parS55 and the HpSpo0J-parS complex is formed. In addition, the ATPase activities of HpSoj were detected by HpSpo0JN (the N-terminal domain only of HpSpo0J) with 3.7- and 3.1-fold stimulation in parS55 and 55DNA, respectively (Figure 1A). Although HpSpo0JN does not contain the DNA binding domain it still can stimulate the ATPase activity of HpSoj. It suggested that HpSpo0JN enhances ATP hydrolysis of HpSoj. In addition, we conducted the MST experiment to detect the potential interaction between HpSoj and HpSpo0JN and a weak interaction was observed (Figure 1B).

Figure 1.

HpSoj characterization. (A) ATP hydrolysis. The purified HpSoj derivatives were incubated with ATP in reaction mixtures containing 55DNA, parS55, HpSpo0J derivatives or both. The released inorganic phosphate (Pi) was detected spectrophotometrically. Non-enzymatic hydrolysis of ATP was undetectable, and the resulting absorbance was subtracted from that of each reaction mixture. Fold stimulation was calculated by setting the ATP hydrolysis rate of HpSoj as one. Error bars represent the standard error of the mean (n = 3). (B) Microscale thermophoresis binding measurements of HpSoj and HpSpo0JN. The labeled HpSpo0JN were mixed with serially diluted HpSoj (•) or lysozyme (▴) which was used as a negative control. (C) EMSA of HpSoj DNA binding. HpSoj binds parS55 in a concentration-dependent fashion in the presence of ATP, but binding is weak at a protein to parS55 molar ratio of 40:1 in the presence of ADP. HMW – high molecular weight.

HpSoj characterization. (A) ATP hydrolysis. The purified HpSoj derivatives were incubated with ATP in reaction mixtures containing 55DNA, parS55, HpSpo0J derivatives or both. The released inorganic phosphate (Pi) was detected spectrophotometrically. Non-enzymatic hydrolysis of ATP was undetectable, and the resulting absorbance was subtracted from that of each reaction mixture. Fold stimulation was calculated by setting the ATP hydrolysis rate of HpSoj as one. Error bars represent the standard error of the mean (n = 3). (B) Microscale thermophoresis binding measurements of HpSoj and HpSpo0JN. The labeled HpSpo0JN were mixed with serially diluted HpSoj (•) or lysozyme (▴) which was used as a negative control. (C) EMSA of HpSoj DNA binding. HpSoj binds parS55 in a concentration-dependent fashion in the presence of ATP, but binding is weak at a protein to parS55 molar ratio of 40:1 in the presence of ADP. HMW – high molecular weight.

We assessed the DNA-binding ability of HpSoj by EMSA using parS24 and parS55 fragments and with or without nucleotide ADP or ATP (Figure 1C and Supplementary Figure S2). We found that the DNA-binding ability of HpSoj is dependent on ATP. The DNA-binding were observed under ATP and ADP, at the concentrations of HpSoj of 0.4 and 1.5 mg/ml, respectively (Figure 1C and Supplementary Figure S2). Meanwhile, HpSoj completely bound DNA at a molar ratio of 40:1 in the presence of ATP (Figure 1C).

Overall structure of HpSoj

HpSoj exhibits Rossmann folding, which is a typical of nucleotide-binding protein (Figure 2A), and forms a dimer (Figure 2B). HpSoj comprises a central seven-stranded β-sheet (β1–β7) framed by α1–α2 and α9–α11 on one side and α3-α8 on the other side (Figure 2A). The ATP molecule in complex with the hexa-coordinated magnesium ion was wedged into a surface-exposed cavity formed by helices α1 and α9 and loops β1α1 and α8α9 (Figure 2A). Each HpSoj monomer binds to one ATP that is sandwiched between the interfaces of the dimer (Figure 2B). The ATP is bound to loop β1α1 (residues 11–19) of the conserved Walker A motif, and is shielded from the other monomer by loop β6α7 (residues 155–158) (Figure 2B). However, the highly conserved Lys12 residue of the neighboring monomer can interact with the ATP by extended hydrogen bonding (Supplementary Table S1). In addition, Gly13 and Gly14 from the neighboring monomer make van der Waal contacts with the ATP molecule (Supplementary Table S1). HpSoj forms a tight dimer with a buried surface area of 1440 Å2, as calculated by PISA (48) (Figure 2B). Dimerization of HpSoj involves seven hydrogen bonds and numerous hydrophobic interactions (Supplementary Table S2). Based on the crystal structures of ParA superfamily, there are also several hydrogen bonds and hydrophobic interactions exited in TtSoj, SpParA and TP228 ParA dimer interface. However, the conservation among these dimerization residues is low.

Figure 2.

Crystal structures of HpSoj-ATP and its DNA complexes. (A) HpSoj monomer. The structure of HpSoj-ATP monomer comprises eleven α-helices (α1- α11) and seven β-strands (β1–β7). ATP is shown as a stick and the magnesium ion is shown as a yellow sphere. (B) HpSoj dimer. The two monomers of the dimer are colored in magenta and cyan, respectively. The monomer colored in magenta is rotated 90° along the y-axis relative to the monomer shown in (A). (C) The ATP-binding site of the HpSoj-ATP complex. (D) The HpSoj–ATP–DNA complex. The overall structure of the HpSoj–ATP–DNA complex is shown as a ribbon model. The monomers of the dimer are colored in magenta or cyan, respectively. The DNA molecule bound to the two dimers is colored pale orange. The ATP-binding sites of (E) the HpSoj–ATP–DNA complex and (F) the HpSoj–ADP–DNA complex. In (C), (E) and (F), the Fo– Fc omit electron density maps of ATP and ADP are contoured at 3.0σ and shown as a mesh. The ATP/ADP-interaction residues from two monomers of the dimer are shown as sticks and are colored magenta and cyan, respectively. Water molecules in the ATP-binding pocket are shown as red spheres, and the nucleophilic water molecule is labeled as WNu. The magnesium ion is shown as a yellow-green sphere. Hydrogen-bond interactions are shown as dashed lines.

Crystal structures of HpSoj-ATP and its DNA complexes. (A) HpSoj monomer. The structure of HpSoj-ATP monomer comprises eleven α-helices (α1- α11) and seven β-strands (β1–β7). ATP is shown as a stick and the magnesium ion is shown as a yellow sphere. (B) HpSoj dimer. The two monomers of the dimer are colored in magenta and cyan, respectively. The monomer colored in magenta is rotated 90° along the y-axis relative to the monomer shown in (A). (C) The ATP-binding site of the HpSoj-ATP complex. (D) The HpSoj–ATP–DNA complex. The overall structure of the HpSoj–ATP–DNA complex is shown as a ribbon model. The monomers of the dimer are colored in magenta or cyan, respectively. The DNA molecule bound to the two dimers is colored pale orange. The ATP-binding sites of (E) the HpSoj–ATP–DNA complex and (F) the HpSoj–ADP–DNA complex. In (C), (E) and (F), the Fo– Fc omit electron density maps of ATP and ADP are contoured at 3.0σ and shown as a mesh. The ATP/ADP-interaction residues from two monomers of the dimer are shown as sticks and are colored magenta and cyan, respectively. Water molecules in the ATP-binding pocket are shown as red spheres, and the nucleophilic water molecule is labeled as WNu. The magnesium ion is shown as a yellow-green sphere. Hydrogen-bond interactions are shown as dashed lines.

We further characterized the ATP molecule in the HpSoj-ATP structure by the Fo – Fc omit map (Figure 2C), and found that its binding pocket mainly comprises α1 and loop β1α1 (Walker A motif), loop β2α2 (Walker A’ motif) and β5 (Walker B motif), as well as β6, α9 and loops β5α6, β6α7, α8α9 (Figure 2A and Supplementary Figure S1). The ATP is completely trapped between the interfaces of the HpSoj dimer by numerous interactions (Figure 2C and Supplementary Table S1). The magnesium ion is hexa-coordinated with Thr18, ATP O1B, ATP O3G, and three water molecules (Supplementary Table S1c).

According to previous reports on Soj (7,10), a key functional aspartic acid residue (Asp41 in HpSoj) co-coordinates the water nucleophile (WNu) and initializes ATP hydrolysis, playing an important role in ATPase activity. Asp41 of HpSoj binds the WNu and interacts with the γ-phosphate of ATP to initiate ATP catalysis (Figure 2C), as found for Walker-type ATPases. The ATP-binding residues are all conserved and critical in the ParA superfamily (7,16), and corresponding residues in HpSoj are 14GVGKTT19 (the ‘deviant’ Walker A motif), Asp41, and Asp130 (Figure 2C). The ATP hydrolysis of HpSojD41A was reduced to 0.4 ± 0.2 mol Pi released/mole HpSojD41A·hour−1. The P1 ParAD152N and TtSojD44A (corresponding to HpSojD41A) also have been reported and TtSojD44A mutant was suggested as an ATP-hydrolysis deficient mutant (7,49).

HpSoj–DNA complexes

The overall structure of the HpSoj–ATP–DNA complex is shown in Figure 2D, and the ATP-binding site is detailed in Figure 2E. Both monomers of the HpSoj dimer equally contribute to DNA-binding, reinforcing the essential function of Soj dimerization. Each monomer binds DNA via the α8, α9 and α10 helices, with the DNA located in a furrow-shaped basic patch formed by four key lysine residues: Lys199 (in α8), Lys227 (in α9), Lys230 (in α9) and Lys247 (in α10) (Figure 3A). Among these DNA-binding residues, Lys227, Lys230 and Lys247 are conserved as either Lys/Arg in the bacterial ParA superfamily, and only Lys247 is threonine in CcParA (Supplementary Figure S1). Lys199 reveals sequence diversity as K/I/Q/E in the ParA superfamily (Supplementary Figure S1).

Figure 3.

The HpSoj–ATP–DNA complex. (A) The detailed interactions between HpSoj dimer and DNA of the HpSoj–ATP–DNA complex. A close-up view of the interactions between HpSoj and DNA related to Figure 2d is shown. The residues of the DNA binding environment (N196, K199, K227, V229, K230, and K247) are shown as sticks. The α8, α9, and loop α8α9 features that constitute the minor groove clamp of each monomer of the dimer are colored pink and cyan, respectively. (B) The correlated positions of K227, ATP, and DNA of the HpSoj–ATP–DNA complex. Loop α8α9 and K227 from each monomer of the dimer are colored pink and cyan, respectively. (C) The equilibrium DNA binding assays of HpSoj and its mutant proteins with parS24 DNA were done by fluorescence polarization binding isotherms (FP). The parS24 binding for HpSoj (•), HpSojK199E (▪), HpSojK230E (✦), HpSoj K199/230E (▴) and HpSojK199/227/230/247E (▾) were shown. The corresponding Kd are 122±14 nM (R2 = 0.99), 215±24 nM (R2 = 0.99), 187±36 nM (R2 = 0.96), 308±40 nM (R2 = 0.98) and NB (no measurable binding), respectively. DNA binding was measured by fluorescence polarization (in mFP units), plotted against protein concentration (0–10 μM). The average of three independent experiments is shown with error bars representing one standard deviation of the mean. (D) Superimposition of the HpSoj-ATP and HpSoj–ATP–DNA structures. The α-helices α1, α4, α7 and α9 of HpSoj-ATP and HpSoj–ATP–DNA are colored pink and cyan, respectively. (E) Dimer-dimer interaction regions of the HpSoj–DNA complex. Three regions participating in the dimer-dimer interface are labeled I, II and III and colored in yellow, pink, and green, respectively. Residues involved in the dimer-dimer interactions are labeled.

The HpSoj–ATP–DNA complex. (A) The detailed interactions between HpSoj dimer and DNA of the HpSoj–ATP–DNA complex. A close-up view of the interactions between HpSoj and DNA related to Figure 2d is shown. The residues of the DNA binding environment (N196, K199, K227, V229, K230, and K247) are shown as sticks. The α8, α9, and loop α8α9 features that constitute the minor groove clamp of each monomer of the dimer are colored pink and cyan, respectively. (B) The correlated positions of K227, ATP, and DNA of the HpSoj–ATP–DNA complex. Loop α8α9 and K227 from each monomer of the dimer are colored pink and cyan, respectively. (C) The equilibrium DNA binding assays of HpSoj and its mutant proteins with parS24 DNA were done by fluorescence polarization binding isotherms (FP). The parS24 binding for HpSoj (•), HpSojK199E (▪), HpSojK230E (✦), HpSoj K199/230E (▴) and HpSojK199/227/230/247E (▾) were shown. The corresponding Kd are 122±14 nM (R2 = 0.99), 215±24 nM (R2 = 0.99), 187±36 nM (R2 = 0.96), 308±40 nM (R2 = 0.98) and NB (no measurable binding), respectively. DNA binding was measured by fluorescence polarization (in mFP units), plotted against protein concentration (0–10 μM). The average of three independent experiments is shown with error bars representing one standard deviation of the mean. (D) Superimposition of the HpSoj-ATP and HpSoj–ATP–DNA structures. The α-helices α1, α4, α7 and α9 of HpSoj-ATP and HpSoj–ATP–DNA are colored pink and cyan, respectively. (E) Dimer-dimer interaction regions of the HpSoj–DNA complex. Three regions participating in the dimer-dimer interface are labeled I, II and III and colored in yellow, pink, and green, respectively. Residues involved in the dimer-dimer interactions are labeled.

Residues Asn196, Lys199, and Val229 directly interact via hydrogen bonding with phosphate groups of the DNA backbone in the minor groove (Figure 3A). Surprisingly, these DNA-binding residues are not conserved among the ParA superfamily (Supplementary Figure S1). Residues Asn196 and Val229 interact with DNA through the amine nitrogen of the main chain, whereas Lys199 interacts with it via the amine group of the side chain. These interactions between Soj protein and nsDNA in the minor groove are a common feature shared with other nsDNA-binding proteins (50,51). Furthermore, the continuous basic patch created by Lys199, Lys227, Lys230 and Lys247 in the HpSoj dimer is required for nsDNA-binding. Our sequence alignment (Supplementary Figure S1) reveals that these basic residues are partially conserved as either Lys/Arg in the bacterial ParA superfamily, so a continuous basic patch might be necessary for nsDNA binding.

The crystal structure of the HpSoj–ADP–DNA complex was determined at a resolution of 2.5 Å. The ADP was located in the nucleotide binding site according to the Fo – Fc omit map at 3.0 sigma (Figure 2F). Although ATP had been added in the crystallization setup, the hydrolyzed ADP was observed in the crystal (Figure 2F). Two HpSoj-ADP dimers were bound to one DNA molecule, similar to the structure of the HpSoj–ATP–DNA complex (Supplementary Figure S3A). The overall structures and ATP-binding sites of the HpSoj–ATP–DNA and HpSoj–ADP–DNA complexes are similar to those of HpSoj (Figure 2C, E, F and Supplementary Figure S3). Based on structural comparison, the root mean square deviation (r.m.s.d.) between the HpSoj-ATP and HpSoj–ATP–DNA complexes is 0.48 Å (in Cα), between HpSoj-ATP and HpSoj–ADP–DNA it is 0.51 Å (in Cα), and between HpSoj–ATP–DNA and HpSoj–ADP–DNA it is 0.35 Å (in Cα). Previous study has shown that the average r.m.s.d. for proteins recognizing nsDNA is <1.0 Å between unbound- and bound-DNA conformations (52). The overall B-factors of HpSoj–ADP–DNA and HpSoj–ATP–DNA are 59 and 85 Å2, respectively (Table 1). The observation that both HpSoj–ADP–DNA and HpSoj–ATP–DNA share similar structures may be due to ATP hydrolysis during crystallization, which could prevent any subsequent conformational change that might occur in the ADP-bound form.

Structural differences among HpSoj and its DNA complexes

Structural comparison of HpSoj-ATP and its DNA complexes (Figure 3D) revealed that they share a similar overall structure, with r.m.s.d. values less than 1 Å. PyMOL (53) calculations revealed that helix α7 rotates 5.1° and participates in dimer interactions via residues Glu161/Leu165 (Supplementary Table S2), and that loop β6α7 is involved in ATP binding through residues Glu156/Phe157/Phe158 (Figure 2C and E). Also, we found that helix α9 rotates 4.6° and is involved in DNA binding (Figure 3B), and that loop α8α9 is involved in ATP binding (Figure 3B). The dimer interfaces of the HpSoj-ATP, HpSoj–ATP–DNA and HpSoj–ADP–DNA complexes are 1440, 1349, and 1313 Å2, respectively, based on PISA calculation (48). Given the disparity in interface area between DNA-unbound and -bound HpSoj dimer, the monomer-monomer relationship may have to be adjusted in order to interact properly with DNA. Our PyMOL calculations (53) revealed a difference of 3.6° in the angle between α4 of the dimer for HpSoj–ATP and HpSoj–ATP–DNA. Several hydrogen bonds present in the HpSoj-ATP dimer interaction were absent in the HpSoj–ATP–DNA complex, all of which were located in α2 and α7 including the hydrogen bonds of Gln43-Gln11 and Ser48-Glu161 (Supplementary Table S2).

Non-specific DNA binding

Protein can non-specifically bind with DNA using positively-charged and surface-exposed residues to interact with the negatively-charged phosphate of the DNA backbone (17). HpSoj dimer binds DNA through a continuous basic binding patch formed by Lys199, Lys227, Lys230 and Lys247. Based on the sequence alignment and the structural comparisons, the four corresponding basic residues in TtSoj are Arg182/Arg194/Arg208/Arg211, in BsSoj they are Arg189/Lys201/Arg215/Arg218, and in SpParA they are Lys230/Arg253/Lys256/Lys273 (Supplementary Figure S1). Mutagenesis of BsSoj revealed that Arg189 and Arg218 are involved in DNA-binding and mediate the interaction between BsSoj and the nucleoid (17). In SpParA, Lys256A and Lys273/274A mutation abrogated DNA-binding ability (10,54). Thus, the continuous basic patch may be important for DNA-binding among the bacterial ParA superfamily.

Based on our HpSoj–ATP–DNA complex structure, HpSoj dimer binds nsDNA through a continuous basic binding patch formed by residues Lys199, Lys227, Lys230, Lys247, as well as Asn196, and Val229 (Figure 3A). Lys199/Asn196 and Lys230/Val229 interact with DNA by clamping the minor groove. Asn196 and Lys199 from α8 exhibit low sequence homology among bacterial ParA (Supplementary Figure S1). Asn196 interacts with DNA through the amine nitrogen of the main chain, whereas Lys199 interacts with DNA via the amine group of the side chain. Val229 and Lys230 of α9 are conserved among bacterial ParA (229V-K/R-L/V-A/S-E233) (Supplementary Figure S1). Val229 interacts with DNA through the amine nitrogen of the main chain, but Lys230 does not directly interact with DNA. Lys199 and Lys230 promote DNA binding by inducing their neighboring residues, Asn196 and Val229 respectively, to form hydrogen bonds via their main chains with DNA in the center of the DNA minor groove (Figure 3A). Loop α8α9 exhibits sequence consensus among bacterial ParA (224M/V-I-P-K/R-S/N228) (Supplementary Figure S1). Pro226 and Ser228 directly interact with the adenosine of ATP through their main chains (Figure 2C and Supplementary Table S1). These loop α8α9 interactions with ATP have also been observed in TtSoj (corresponding to residues Pro207/Arg208/Asn209) (7) and SpParA (corresponding to residues Lys252/Arg253/Ser254) (10).

Lys227 is highly conserved in the ParA superfamily (Supplementary Figure S1). Interestingly, Lys227 is located at the core of the DNA binding surface, but its side chain is oriented away from the DNA and it does not directly interact with DNA (Figure 3A). Moreover, this residue is embedded inside the dimer interface and interacts with the ribose of ATP through a water molecule (Figure 2C and Supplementary Table S1). Thus, significantly, Lys227 might play an important role in linking ATP- and DNA-binding in the HpSoj dimer (Figure 3B). The orientation of the Lys227 side chain is important for adjusting loop α8α9 to interact with ATP. In the presence of DNA, Lys227 is located at the core of the HpSoj DNA-binding surface, at a distance of 4.3–7.2 Å from the DNA molecule. In addition, Lys247 exhibits sequence homology among the bacterial ParA superfamily (except for TtSoj: Thr228) (Supplementary Figure S1), and is located 4.9–8.0 Å from DNA, so it might play an interactive role in the DNA binding of bacterial ParA. Thus, based on our analyses of these residues, α8, loop α8α9 and α9 are important functional regions for ATP- and DNA-binding and they link these two functions in the ParA superfamily.

Based on the crystal structure of the HpSoj–ATP–DNA complex (Figure 3A), four key lysine residues (Lys199, Lys227, Lys230 and Lys247) are involved in the non-specific DNA binding. The HpSoj mutants, K199E, K230E, K199/230E and K199/227/230/247E, have been isolated and purified. Meanwhile, their non-specific DNA binding were measured by the fluorescence polarization (FP) and shown in Figure 3C. The Kd of non-specific DNA binding of wildtype HpSoj is calculated to be 122 nM. However, the non-specific DNA binding of the mutants are decreased and the Kd for K199E, K230E and K199/230E are 187, 215, 308 nM, respectively. Nevertheless, the Kd of mutant K199/230/230/247E is undetectable. Indeed, it suggested that Lys199, Lys227, Lys230 and Lys247 play an important role in the non-specific DNA binding of HpSoj.

Protein-protein interactions of HpSoj–DNA complexes

We observed HpSoj dimer–dimer interactions at several locations, but the DNA-binding region is preserved in the DNA-unbound state (Figure 3E). Dimer–dimer interactions were observed in three regions of HpSoj–DNA complexes (I, II and III in Figure 3E). The buried surface areas of these regions were calculated for HpSoj–ATP–DNA and HpSoj–ADP–DNA as 295 and 319 Å2 (region I), 392 and 558 Å2 (region II), and 317 and 305 Å2 (region III), respectively. Among the three interfaces, region II contributes most to protein-protein interactions. Region II is mainly formed by helix α4 that displays diverse sequence and length among the ParA superfamily (Supplementary Figure S1). The calculated buried surface area of region II is 488 Å2 for HpSoj-ATP alone. After DNA binding, HpSoj–ATP–DNA adopts an active state with fewer protein-protein interactions. HpSoj–ADP–DNA may adopt a relaxed state and return to the DNA-unbound state, so it has a similar protein-protein interaction area as that of HpSoj-ATP.

Superimposition of HpSoj-ATP and the HpSoj–ATP–DNA complexes relative to the AB dimer (Supplementary Figure S3B) revealed an ∼9–10 Å deviation between the CD dimers (Cα of Thr98 in α4, and Cα of Lys207 in α8). After DNA binding, dimer-dimer interactions affect the relationship of the AB and CD dimers to induce a molecular shift in the CD dimer (Supplementary Figure S3B). Although the HpSoj dimer only has one DNA-binding site, it possesses numerous dimer-dimer interaction regions to recruit HpSoj molecules (Figure 3E and Supplementary Figure S3B).

The protein-protein interactions of HpSoj dimer might play an essential role in facilitating HpSoj localization to high-density DNA regions (HDRs, Supplementary Figure S4A) in the H. pylori ParABS system. It has previously been reported that the DNA-binding activity of ParA is essential for localization of all components of the partitioning machinery to HDRs and for a stable HDR-bound ParA state (32). Accordingly, the HpSoj–ATP–DNA structure might mimic the stable and active HDR-bound state (Supplementary Figure S4A). In the multifaceted DNA-binding site of pNOB8 ParA-DNA, each ParA dimer is surrounded by a dense DNA substrate (Supplementary Figure S4B) (37). The sole DNA-binding site in our HpSoj–DNA complex structure (Figure 3E), HpSoj dimers can be recruited by protein-protein interactions to engage several DNA substrates (Supplementary Figure S4A).

Structural comparison between HpSoj and pNOB8 ParA

We superimposed the structures of HpSoj and pNOB8 ParA monomers and found an r.m.s.d. of 5.5 Å (Cα atom), with conformational differences at α4 and loop α8α9, as well as several insertions in α8, α9, α10, loop α2α3, loop β3β4, loop α7β7 and loop α10α11 of pNOB8 ParA (Figure 4A). However, HpSoj and pNOB8 ParA share a similar ATP-binding site. The dimer interfaces of HpSoj-ATP and the HpSoj–ATP–DNA complex are 1440 and 1349 Å2, respectively, whereas they are 1861 and 1849 Å2 for pNOB8 ParA and pNOB8 ParA–AMPPNP–DNA. Moreover, the HpSoj and pNOB8 ParA dimers are dissimilar and the relationships between their monomers are different (Figure 4B).

Figure 4.

HpSoj and pNOB8 ParA. (A) Superimposition of the HpSoj, TtSoj, and pNOB8 ParA monomers. The main structural differences are labeled as loop α2α3, α4, and loop α8α9, which are colored green, brown and yellow for HpSoj, TtSoj, and pNOB8 ParA, respectively. (B) Structural superimposition of the HpSoj and pNOB8 ParA dimers. The HpSoj and pNOB8 ParA dimers are colored green and yellow, respectively. (C) Structural superimposition of the HpSoj–DNA and pNOB8 ParA-DNA dimers. The orientation of HpSoj–DNA is rotated 90° along the y-axis of that complex in (B). HpSoj and pNOB8 ParA are colored as shown in (B) and their bound DNAs are shown as ribbons and colored accordingly. (D) Electrostatic surface representation of the HpSoj–DNA complex. Positive and negative potentials are shown in blue and red, respectively. The orientation of the HpSoj–DNA complex is rotated 90° along the x-axis relative to that in (B). Residues involved in DNA-binding from each monomer of the dimer are labeled. Residues Lys199, Lys227, Lys230 and Lys247 that form the basic-patch for DNA binding are labeled and underlined. (E) Electrostatic surface representation of the pNOB8 ParA-DNA complex. Positive and negative potentials are shown in blue and red, respectively. The orientation of the pNOB8 ParA–DNA complex is rotated 90° along the x-axis relative to that in (C).

HpSoj and pNOB8 ParA. (A) Superimposition of the HpSoj, TtSoj, and pNOB8 ParA monomers. The main structural differences are labeled as loop α2α3, α4, and loop α8α9, which are colored green, brown and yellow for HpSoj, TtSoj, and pNOB8 ParA, respectively. (B) Structural superimposition of the HpSoj and pNOB8 ParA dimers. The HpSoj and pNOB8 ParA dimers are colored green and yellow, respectively. (C) Structural superimposition of the HpSoj–DNA and pNOB8 ParA-DNA dimers. The orientation of HpSoj–DNA is rotated 90° along the y-axis of that complex in (B). HpSoj and pNOB8 ParA are colored as shown in (B) and their bound DNAs are shown as ribbons and colored accordingly. (D) Electrostatic surface representation of the HpSoj–DNA complex. Positive and negative potentials are shown in blue and red, respectively. The orientation of the HpSoj–DNA complex is rotated 90° along the x-axis relative to that in (B). Residues involved in DNA-binding from each monomer of the dimer are labeled. Residues Lys199, Lys227, Lys230 and Lys247 that form the basic-patch for DNA binding are labeled and underlined. (E) Electrostatic surface representation of the pNOB8 ParA-DNA complex. Positive and negative potentials are shown in blue and red, respectively. The orientation of the pNOB8 ParA–DNA complex is rotated 90° along the x-axis relative to that in (C).

In the HpSoj–DNA complex, the continuous basic patch of HpSoj is exposed for DNA binding (Figure 4D). The pNOB8 ParA and its DNA complex reveal a different dimer conformation (35,37). The corresponding DNA-binding residues of HpSoj (Asn196, Lys199, Lys227, Val229, Lys230, Lys247) in pNOB8 are Arg210/Ala214/Arg257/Lys258/Glu259/Val288, but Arg257/Lys258/Glu259/Val288 are missing from the crystal structure of the ParA–AMPPNP–DNA complex (PDB ID: 5U1J). In the pNOB8 ParA–AMPPNP–DNA complex, the two monomers of the ParA dimer need to be rotated to expose basic residues (Arg52, Lys58, Lys85, Lys218, Lys221 and Lys 270) for optimal DNA binding. HpSoj region III (responsible for dimer-dimer interactions; Figures 3E and 4C) corresponds to the DNA-binding residues of pNOB8 ParA. Notably, the electrostatic surface distribution of the DNA-binding region of HpSoj is predominantly basic, whereas it is evenly distributed among basic/acidic in the pNOB8 ParA–AMPPNP–DNA complex (PDB ID: 5K5Z) (Figure 4D and E).

The bacterial and archaeal partitioning system

The partitioning system in most bacteria involves two proteins (ParA and ParB) and a cis-acting centromere-like parS DNA (55). Two partitioning systems have been described in archaea; the SegAB system and AspA–ParBA that comprise different key functional proteins (36). We superimposed the monomer structures of HpSoj and TtSoj (Figure 4A) and found that they share a similar overall structure, with an r.m.s.d. of 0.8 Å (Cα atom). Structural differences between the HpSoj and TtSoj monomers mainly occur in α4 and loops α2α3 and α8α9 (Figure 4A). Loop α2α3 and α4 are involved in dimer-dimer interactions (Supplementary Table S2), and loop α8α9 participates in ATP-binding (Supplementary Table S1). HpSoj has longer α4 (residues 99–111) and loop α8α9 (residues 211–224) regions than TtSoj (Supplementary Figure S1). However, the dimer relationships of both HpSoj and TtSoj are quite similar (Figure 5A and B). Based on the crystal structure of the HpSoj–ATP–DNA complex, we postulate a possible DNA-binding patch in TtSoj created by residues Arg182, Arg208, Arg211 and Thr228 (Figure 5B). Significantly, a basic patch can also be observed for SpParA (Lys230, Arg253, Lys256 and Lys273) (Figure 5C) and for TP228 ParA (Asn148, Arg169, Lys174 and Lys191) (Figure 5D), which also correspond to the DNA-binding basic patch of HpSoj. These findings suggest that bacterial ParA superfamily might have a similar nsDNA-binding mode as that HpSoj has.

Figure 5.

Structural comparison of the bacterial ParA superfamily. Comparison of the electron surface potentials of various bacterial ParA proteins. The HpSoj, TtSoj, SpParA and TP228 ParA dimers are shown as ribbons (left panel), and their electronic surface potentials (right panel) are shown as 90° rotations along the x-axis of the view shown in the left panel. Positive and negative potentials are shown in blue and red, respectively.

Structural comparison of the bacterial ParA superfamily. Comparison of the electron surface potentials of various bacterial ParA proteins. The HpSoj, TtSoj, SpParA and TP228 ParA dimers are shown as ribbons (left panel), and their electronic surface potentials (right panel) are shown as 90° rotations along the x-axis of the view shown in the left panel. Positive and negative potentials are shown in blue and red, respectively.

In the archaeal SegAB system, the ATPase SegA is polymerized upon binding ATP, and the site-specific DNA-binding protein SegB promotes SegA polymerization (34). In the archaeal AspA-ParBA machinery, pNOB8 ParA–AMPPNP forms a polymer in the absence of DNA but not in the presence of DNA (37). HpSoj does not form a polymer, regardless of the presence of nucleotide or DNA. HpSoj and pNOB8 ParA reveals different monomer relationships of their dimers (Figure 4B). Remarkably, there is a significant discrepancy in DNA binding between HpSoj and pNOB8 ParA (Figure 4C); HpSoj dimer binds one DNA at the base of the molecule, whereas the pNOB8 ParA dimer binds two DNA molecules at two sides of the molecule (Figure 4C). HpSoj and pNOB8 ParA also have different DNA-binding surfaces; 504 Å2 for the sole DNA-binding site of HpSoj and 238 and 194 Å2 for the two pNOB8 ParA DNA-binding sites (Figure 4C), as calculated by PISA (48). The solvent content of the HpSoj–ATP–DNA and pNOB8 ParA–AMPPNP–DNA complexes are 58% and 45%, respectively. The pNOB8 ParA–AMPPNP–DNA complex contains multiple DNA-binding sites and the low solvent content induces tight molecular packing between pNOB8 ParA and the DNA (Supplementary Figure S4B). Thus, structural comparison of HpSoj–ATP–DNA and pNOB8 ParA–AMPPNP–DNA complexes demonstrates that the mode of nsDNA-binding might differ between bacteria and archaea. Despite this difference, they might share a similar DNA segregation mechanism to drive movement of the ParB–DNA cargo, which does not require ParA polymerization.

Nucleoid–adaptor complex (NAC) formation

To investigate ParABS nucleoid-adaptor complex (NAC) formation in H. pylori, we used EM to directly capture HpSpo0J and HpSoj in complex with DNA (Figure 6). We used parS550 for specific DNA-binding by HpSpo0J and 180DNA as a substrate for HpSoj to probe the HpSpo0J–HpSoj–DNA complex based on the differing DNA lengths. We assessed [HpSpo0J:parS550] experiment, a total of 203 conformational complexes were counted, including 70.9% of the single HpSpo0J:parS550 complex (Figure 6A), 12.8% of the HpSpo0J:parS550 paired complex (Figure 6B), and 16.3% of a higher-order complex (Figure 6C) (Table 2). However we observed NAC complex formation when we co-incubated [HpSpo0J:parS550] (the partitioning complex, cargo) with [HpSoj–180DNA] (the nucleoid-bound complex) (Figure 6F–H). We counted a total of 173 conformational complexes in this latter experiment (Table 2), comprising 72.3% of the single HpSpo0J:parS550 complex, 6.9% of the HpSpo0J:parS550 paired complex, 18.5% of the [HpSpo0J:parS550]+[HpSoj:180DNA] complex (i.e. the NAC complex), and 2.3% of a higher-order complex. Thus, significant amounts of NAC complexes were detected when both HpSoj and HpSpo0J interacted with the assistance of the specific parS DNA. It suggests that HpSoj and DNA involvements might promote HpSoj–HpSpo0J interaction to form the NAC complex, instead of making HpSpo0J–HpSpo0J interaction to form the high order complex.

Figure 6.

Electron microscopy images of HpSoj and HpSpo0J nucleoprotein complexes. EM images of HpSoj, HpSpo0J and DNA complexes. (A) The HpSpo0J–parS550 single complex. (B) The HpSpo0J–parS550 paired complex. (C) The higher-order complex of HpSpo0J-parS550. (D) The HpSoj-180DNA complex. (E) The HpSoj-180DNA paired complex. (F), (G) and (H) are images of three different [HpSpo0J:parS550]+[HpSoj:180DNA] complexes.

Electron microscopy images of HpSoj and HpSpo0J nucleoprotein complexes. EM images of HpSoj, HpSpo0J and DNA complexes. (A) The HpSpo0J–parS550 single complex. (B) The HpSpo0J–parS550 paired complex. (C) The higher-order complex of HpSpo0J-parS550. (D) The HpSoj-180DNA complex. (E) The HpSoj-180DNA paired complex. (F), (G) and (H) are images of three different [HpSpo0J:parS550]+[HpSoj:180DNA] complexes.

The [HpSpo0J:parS550]+[HpSpo0J:180DNA] data revealed that DNA with a specific parS sequence can facilitate HpSpo0J–HpSpo0J interaction to form the HpSpo0J-DNA paired complex and the higher-order complexes, whereas the DNA without the parS site might disrupt formation of these complexes. Also, we used DNA without a parS site (600DNA) to examine formation of the HpSpo0J-DNA paired complex. HpSpo0J:600DNA paired and higher-order complexes accounted for 3.6% and 0.5%, respectively (Supplementary Table S3) (as opposed to 12.8%/16.3% in HpSpo0J:parS550 complexes), suggesting that parS DNA is essential for their formation. Moreover, only 2.0% NAC formation was observed for this experiment (Supplementary Table S3) (as opposed to 18.5% for [HpSpo0J:parS550]+[HpSoj:180DNA]), supporting that parS DNA promotes NAC formation. Furthermore, inclusion of HpSoj disrupted formation of the HpSpo0J:600DNA paired complex (from 3.6% to 1.5%).

DISCUSSION

In this study, we determined the ATPase activity of HpSoj and found that it is promoted by HpSpo0J and DNA-binding. We solved the crystal structures of HpSoj and its DNA-bound complexes, HpSoj–ATP–DNA and HpSoj–ADP–DNA. From our EM data, we observed possible interactions among HpSoj, HpSpo0J, and DNA and propose potential NAC formation for the H. pylori ParABS system.

HpSoj behaves typically as other ParAs revealing a weak ATPase that is stimulated by DNA and HpSpo0J/ParB. The ATPase activity of HpSoj is stimulated by the HpSpo0J-parS complex. When HpSpo0J binds to parS55 and the HpSpo0J-parS complex is formed the N-terminal domain of HpSpo0J might be exposed based on the crystal structure of the HpSpo0J-parS complex (2). And then the N-terminal domain might interact with HpSoj to promote the ATPase activity. In addition, the ATPase activity of HpSoj was stimulated by HpSpo0JN in the presence of DNA. Although HpSpo0JN does not contain the DNA binding domain it still can stimulate the ATPase activity. This phenomenon also has been reported that the SopB ATPase activation domain exists at its N-terminus, which does not have the DNA binding domain either (56). Several previous functional studies demonstrated that the N-terminal domain of ParB likely contributes to the ParA/ParB interaction and to promote ParA ATPase activity (7,16,57,58). A weak interaction between HpSoj and HpSpo0JN was detected by MST and it perhaps due to their dynamic and transient association.

Single-molecule and 3D structured illumination microscopy studies (17,18,31,32) have indicated that the nsDNA-binding ability of ParA is critical in DNA segregation. The HpSoj dimer binds to one DNA molecule, whereas the archaeal Sulfolobus pNOB8 ParA dimer binds to two DNA molecules via multifaceted interactions (37) (Figure 4C). The DNA-binding regions of HpSoj and pNOB8 ParA differ, involving different non-conserved residues, perhaps because of species variation or because the DNA-binding of ParA is non-specific. Regarding the electrostatic surface of the ParA dimer, a basic patch environment and a proper binding curvature are required for nsDNA binding.

Previous biochemical studies identified that ParA binds with ATP to form a dimer and then undergoes a slow conformational change into the ‘ParA*2-ATP2’ state that is stable and, based on tryptophan fluorescence assays on P1 ParA, precise for DNA binding (26). A continuous and basic DNA-binding patch (comprising residues Lys199, Lys 227, Lys 230 and Lys247) is created when HpSoj is dimerized in the ParA*2-ATP2 state. However, there is no significant conformational difference between HpSoj-ATP and the HpSoj–ATP–DNA complex, as is the case for other nsDNA-binding proteins (52). Since the DNA-binding patch is on the HpSoj dimer surface and readily exposed for DNA binding, a further conformational change is not necessary. In the absence of ATP, HpSoj was not stable and the crystallization was not succeeded. Since the three-dimensional structure of apo-HpSoj (no nucleotide) is not available we cannot rule out the possibility of a different conformation among apo-HpSoj, HpSoj–ATP and HpSoj–ATP–DNA. In addition, multiple conformations of HpSoj-ATP might exist in solution as that reported in P1 ParA (26).

The pNOB8 ParA, which resembles bacterial ParA, possesses nsDNA-binding ability (37), and a multifaceted DNA-binding mode. The pNOB8 ParB contains an N-terminal domain (residues 1–320) that corresponds to bacterial ParB (182–336 residues), albeit with low sequence homology (35). The pNOB8 ParB binds DNA in a non-specific manner; however, bacterial ParB specifically binds DNA at the parS site (35). These Walker-box partitioning systems are ubiquitous and exist in both bacteria (ParABS) and archaea (SegAB and AspA-ParBA), though each exhibits their own particular mechanisms. In contrast, the DNA-binding patch of pNOB8 ParA is buried inside the dimer so monomer rotation is necessary to expose its basic residues for optimal DNA binding (37). Residues Lys199, Lys 227, Lys 230 and Lys247 form the basic DNA-binding patch of HpSoj, with the corresponding residues in pNOB8 ParA being Ala214, Glu250, Ile253 and Lys270. The HpSoj basic residues form a DNA-binding patch that harbors one DNA molecule, whereas the dispersed pNOB8 ParA basic residues result in a multifaceted DNA-binding surface, together suggesting that the DNA-binding mechanism of bacterial ParA is significantly different from that of archaeal ParA.

The HpSoj–ADP–DNA complex presented a similar conformation to that of HpSoj–ATP–DNA. Thus, the HpSoj–ADP–DNA complex might represent a transitional state in which the ATP has been hydrolyzed to ADP but the DNA remains bound. Interaction of HpSoj with HpSpo0J stimulates ATPase activity (Figure 1A) and may accelerate dimer dissociation and DNA release. In Salmonella newport TP228, ParB interacts with the dimer interface of ParA, which may induce ParA to dissociate from DNA, resulting in segregation along the nucleoid matrix (37).

HpSoj does not form a nucleoprotein filament based on our EM data. Also, the hypothesis of ParA polymerization has been excluded with biochemical data and super-resolution microscopy (26,27,29–32). HpSoj may be similar to all ParAs that follow the diffusion-ratchet mechanism whereby the ParB partitioning cargo moves on the nucleoid through the chemical gradient of ParA. However, structural results show that the architectures of the DNA segregation systems in bacteria and archaea exhibit two distinct differences: (i) bacterial ParA have only one nsDNA-binding site, but archaeal ParA have two; and (ii) bacterial ParB specifically use the N-terminus for ParA interaction (7,35,56,57,59–61), whereas archaeal ParB utilizes a flexible linker to interact with ParA (35). Moreover, ParB in archaeal AspA-ParBA systems does not possess specific DNA-binding ability, which instead is carried out by another protein, AspA (35).

Despite significant differences in the sequence homology, in the manner of DNA binding, and in the three-dimensional structures of ParA protein, both bacteria and archaea modulate DNA segregation that does not require ParA polymerization. The ParA/ParB complex may undergo dynamic and transient conformational change to control the movement of the ParB-DNA cargos (28,29). Interaction between ParA and ParB is known to be weak (6,26), as we observed from our MST assay on HpSoj and HpSpo0JN (Figure 1B).

Previous EM analysis revealed ParB-mediated centromere pairing in E. coli (20), with both paired and higher-order (36%) complexes forming at low ParB concentrations, whereas mostly higher-order complexes were formed at high ParB concentrations. That study concluded that ParB mediated centromere pairing through its N-terminal end. The crystal structure of the HpSpo0J and parS complex in H. pylori (2) is an elongated structure, with a flexible N-terminal domain for protein–protein interaction and a conserved DNA-binding domain for parS binding. HpSpo0J interacts vertically and horizontally with its neighbors through the N-terminal domain to form an oligomer by adjacent and transverse interactions that might be necessary for molecular assembly of a higher-order nucleoprotein complex (2).

Our EM results (Figure 6 and Table 2 and Supplementary Table S3) show that the HpSpo0J/ParB paired complex was significantly reduced when the DNA substrated lacked a parS site (600DNA) and when HpSoj/ParA ([HpSoj:180DNA]) was included. Thus, the parS site might regulate the HpSpo0J/ParB pairing interaction. In contrast, participation of HpSoj/ParA promotes the HpSoj-HpSpo0J (ParA-ParB) interaction to make the HpSpo0J–HpSoj–DNA complex with the assistance of the parS site. In summary, NAC complex formation could be promoted through HpSoj participation and the assistance of specific parS site DNA. In E. coli P1 ParABS, protein-DNA binding by the ParA–ParB–DNA complex occurs through both specific and non-specific DNA binding by ParB and ParA, respectively, and parS site stabilizes this nucleoprotein complex formation (22). Similarly, the NAC is created by interaction of plasmid-bound ParB and nucleoid-bound ParA (32). The DNA replication in H. pylori, the ParB paired complex was formed to condense the DNA. When the ParB paired complex encounters ParA, this complex may be destroyed by the ParA and ParB interactions that may promote the NAC formation to carry out DNA segregation.

DATA AVAILABILITY

The atomic coordinates and structure factors have been deposited in the Protein Data Bank (http://www.rcsb.org/pdb) with PDB ID codes 6IUB (HpSoj-ATP), 6IUD (HpSoj–ADP–DNA) and 6IUC (HpSoj–ATP–DNA).

Click here for additional data file.