PglZ from Type I BREX phage defence systems is a metal-dependent nuclease that forms a sub-complex with BrxB

Abstract

BREX (Bacteriophage Exclusion) systems, identified through shared identity with Pgl (Phage Growth Limitation) systems, are a widespread, highly diverse group of phage defence systems found throughout bacteria and archaea. The varied BREX Types harbour multiple protein subunits (between four and eight) and all encode a conserved putative phosphatase, PglZ, and an equally conserved, putative ATPase, BrxC. Almost all BREX systems also contain a site-specific methyltransferase, PglX. Despite having determined the structure and fundamental biophysical and biochemical behaviours of several BREX factors (including the PglX methyltransferase, the BrxL effector, the BrxA DNA-binding protein, and a commonly-associated transcriptional regulator, BrxR), the mechanism by which BREX impedes phage replication remains largely undetermined. In this study, we identified a stable BREX sub-complex of PglZ:BrxB, generated and validated a structural model of that protein complex, and assessed the biochemical activity of PglZ from BREX, revealing it to be a metal-dependent nuclease. PglZ can cleave cyclic oligonucleotides, linear oligonucleotides, plasmid DNA and both non-modified and modified linear phage genomes. PglZ nuclease activity has no obvious role in BREX-dependent methylation, but does contribute to BREX phage defence. BrxB binding does not impact PglZ nuclease activity. These data contribute to our growing understanding of BREX phage defence.

Graphical Abstract

Graphical Abstract.

Introduction

Up to 10% of bacterial and archaeal genes are dedicated to phage defence [1]. The mechanisms employed to defend against phage infection are diverse and include prevention of viral entry, induction of cell dormancy or death upon infection, and mechanisms that degrade viral genomes or block viral DNA replication [2, 3]. Phages combat these systems through the evolution of elaborate countermeasures that block their action, leading to viral resistance and a continuous arms race between phage and bacterial populations [4]. Recent analyses have demonstrated that bacteria encode far more phage defence systems than just the most well-studied ‘first responder’ systems such as restriction endonucleases and CRISPR [5–10]. Furthermore, many of these newly discovered bacterial systems display obvious similarities to human innate viral defense systems, implying common evolutionary origins and related mechanisms of action [11].

Originally discovered in the early 1980s [12], Phage Growth Limitation (Pgl) [13, 14] and evolutionarily related Bacteriophage Exclusion (BREX) systems are widespread in bacterial and archaeal species [15]. BREX systems are encoded by single operons, often within genetic defence islands. They are currently categorised into at least six sub-types based on the number of genes in each system (typically four to eight) and on the sequence-based functional annotation of those individual genes and putative translated protein subunits [15]. Type I systems, the most common sub-type, typically comprise six conserved genes and can readily be assayed for two phenotypes, (i) phage defence and (ii) BREX-dependent methylation. Molecular details of the mechanisms of each process are largely unknown.

Pgl and BREX systems have two genes in common. The first is named pglZ (also known as brxZ) and the second is brxC [15]. The PglZ domain of a two-component signalling system response regulator, PorX, has recently been shown to degrade cyclic nucleotides [16], but any equivalent activity within BREX has not yet been explored. Beyond PglZ and BrxC, most BREX systems include a gene encoding a site-specific methyltransferase, termed PglX (also known as BrxX). The identity and order of the remaining genes in each BREX Type vary significantly: various BREX systems encode protein subunits with domains that display recognisable homology to kinases, phosphatases, DNA and/or nucleotide binding domains, DNA modification enzymes, chambered AAA + ATPases, and/or DNA helicases, as well as proteins of unknown function (such as BrxB). Several BREX subunits are quite large (>100 kDa), with significant regions having unknown structure-function properties and behaviours, flanking domains with well-annotated putative functions. Type I BREX systems, like their related counterparts, do not contain any readily identifiable DNA nuclease domains or subunits and appear to likely restrict phage by inhibiting phage DNA replication within the infected bacterial cell.

Despite having previously determined the high resolution structures and biochemical activities of the PglX methyltransferase [17, 18], BrxR (a WYL-domain helix-turn-helix DNA binding transcriptional regulator) [19, 20], BrxA (a small DNA binding protein) [21], and BrxL (a chambered AAA⁺ ATPase and dsDNA binding protein) [22], the mechanisms by which BREX systems function to restrict phage replication and to protect the host genome from the system's activity is still largely unknown. In this study, we have performed analyses to further characterise BREX function, using several Type I BREX systems we have previously investigated: those from Salmonella Typhimurium [17, 23], Escherichia fergusonii [24] (Fig. 1A) and Acinetobacter [20, 22]. Multiple in cellulo pull-down and co-expression analyses using these systems identified larger BREX complexes and a particularly stable sub-complex formed by PglZ and BrxB. A hybrid structural model of the PglZ:BrxB complex and of the protein-protein interface was generated using a mutually-reinforcing combination of computational predictions via AlphaFold3 [25], single-particle cryo-electron microscopy (cryo-EM) analyses, and mutational analyses. Additional biochemical analyses demonstrated that PglZ recapitulates PorX activity and is a metal-dependent nuclease that can cleave not only a broad range of cyclic and linear oligonucleotides, but also plasmid and linear dsDNA. The BrxB interaction does not impact PglZ nuclease activity, nor is a nuclease required for BREX-dependent methylation. Nuclease activity does, however, contribute to BREX phage defence. These data contribute to our growing understanding of the elusive BREX mechanism.

Figure 1.

BREX proteins form higher order complexes. (A) Schematic of the BREX phage defence islands from Salmonella and Escherichia fergusonii. The Salmonella island encodes BREX and PARIS (ariA, ariB) defence systems. The E. fergusonii island encodes BREX and a Type IV restriction enzyme of the GmrSD family, BrxU. (B) Pull-down of Salmonella BREX complexes. E. coli BL21 (DE3) pRARE was transformed with an inducible plasmid expressing His-strep tagged BrxB (or vector control), and a second plasmid expressing the six BREX genes brxA, brxB, brxC, pglX, pglZ and brxL (or vector control). Pull-down samples were analysed by SDS-PAGE and indicated bands were identified by MS. (C) Expression of Salmonella BREX proteins in pairs from pET DUET-based vectors, pulled down with His-SUMO-BrxB. Pull-down samples were analysed on SDS-PAGE and indicated bands were identified by MS. (D) Toxicity during expression of Salmonella BREX proteins, measured as viable counts. Error bars represent the standard deviation of the mean from triplicate data..

Materials and methods

Bacterial strains and culture conditions

Escherichia coli strains DH5α (Invitrogen), ER2796 (New England Biolabs) [26], ER2566 (New England Biolabs), Rosetta 2 (DE3) pLysS (Novagen), BL21 (DE3) pRARE (Novagen), BL21 (DE3) RIL (Novagen), and T7 Express (New England Biolabs) were routinely grown at 37°C, either on agar plates or shaking at 150 rpm for liquid cultures. 2x Yeast Extract Tryptone (YT) was used as the standard growth media for liquid cultures, and Luria Broth (LB) was supplemented with 0.35% (w/v) or 1.5% (w/v) agar for semi-solid and solid agar plates, respectively. When necessary, growth media was supplemented with ampicillin (Ap, 100 μg/ml), chloramphenicol (Cm, 25 μg/ml), kanamycin (Km, 100 μg/mL), spectinomycin (Sp, 100 μg/mL), and isopropyl-β-D-thiogalactopyranoside (IPTG, 1 mM). Growth was monitored using a spectrophotometer (WPA Biowave C08000) measuring optical density at 600 nm (OD₆₀₀).

DNA isolation and manipulation

Plasmid DNA was purified from transformed DH5α cells using an NEB Monarch® Plasmid MiniPrep kit following the manufacturer’s instructions. Larger amounts of negatively supercoiled plasmid pSG483 [27] DNA for assays was purified from transformed DH5α cells using a Machery-Nagel NucleoBond Xtra Midi Plus EF kit following the manufacturer’s instructions. Plasmid DNA was eluted in MiliQ and stored at −20°C. Plasmids are described in Supplementary Table S1.

Phage genomic DNA was purified by incubating 450 μl phage lysate with 4.5 μl DNase I (1 mg/ml; Sigma-Aldrich) and 12.5 μl RNase A (10 mg/ml; ThermoFisher) for 30 min at 37°C. The lysate was further incubated with 2.25 μl proteinase K (20 mg/ml; Sigma-Aldrich) and 23 μl of 10% (w/v) SDS for 30 min at 37°C. The sample was mixed with 500 μl UltraPure™ phenol:chloroform:isoamyl alcohol (25:24:1; v/v/v) (ThermoFisher) and centrifuged at 16 000 x g for 5 min at 4°C. The aqueous layer was removed and carried forward, and the previous step was repeated. The resulting aqueous layer was mixed with 500 μl chloroform:isoamyl alcohol (24:1; v/v) and centrifuged at 16 000 x g for 5 min at 4°C. The aqueous layer was carried forward and incubated with 45 μl 3 M sodium acetate pH 5.2 and 500 μl isopropanol for 15 min at room temperature, before being centrifuged at 16 000 x g for 20 min and 4°C. The supernatant was removed, and the pellet was washed with 70% ethanol by gentle aspiration before being dried at room temperature. The dry pellet was soaked in 50 μl of MiliQ and incubated overnight at 4°C. The gDNA was analysed on a 0.75% 1x TAE agarose gel by agarose gel electrophoresis, and stored at −20 °C.

Preparation of nicked, linear, and relaxed form pSG483

Linear pSG483 was obtained through incubation of 10 μg pSG483 with 10 units of BamHI-HF® (New England Biolabs) in 1x CutSmart buffer (New England Biolabs) for 1 h at 37°C. The enzyme was deactivated by incubation at 65°C for 10 min. Nicked pSG483 was obtained by incubating 10–50 μg pSG483 with 10 units Nb.Bpu10I (ThermoFisher) in 1x Buffer R (ThermoFisher) for 4 h at 37°C. The reaction was terminated by incubation at 80°C for 20 min.

For production of relaxed pSG483, 50 μg nicked pSG483 was further incubated with 1 mM ATP and 10 units T4 ligase (New England Biolabs) for 16 h at room temperature. After ligation, an equal volume of UltraPure™ phenol:chloroform:isoamyl alcohol (25:24:1; v/v/v) (ThermoFisher) was added to the reaction mixture, vortexed briefly, and centrifuged at 16 000 x g for 2 min. The resulting aqueous layer was removed and carried forward. An equal volume of chloroform was added to the aqueous layer before centrifugation at 16 000 x g for 2 min. The resulting aqueous layer was carried forward and 1/10 volume of 3 M sodium acetate pH 5.2 was added, followed by two volumes of 100% ethanol. The sample was mixed by pipetting and stored at −80 °C for 30 min. The sample was centrifuged at 16 000 x g for 20 min at 4°C. The ethanol was removed, and the DNA pellet dried at room temperature. The DNA pellet was resuspended to 300 ng/μl with MiliQ. All DNA products were analysed by agarose gel electrophoresis prior to storage at −20 °C.

Bacterial growth assays

T7 Express E. coli cells were transformed with: empty vectors pETDuet, pCDFDuet, and pCOLADuet (i), pTRB710 (His-SUMO-BrxB, and PglZ (BZ)) (ii), pTRB759 (BrxC and PglX (CX)) (iii), pTRB758 (BrxA and BrxL (AL)) (iv) and combinations of BZ/CX (v), BZ/AL (vi), CX/AL (vii), and BZ/CX/AL (viii). Colonies were inoculated and grown overnight in 5 ml 2x YT with respective antibiotics at 37°C shaking at 180 rpm. Cultures were re-seeded 1:100 (v/v) in 100 ml 2x YT with the relevant antibiotics and grown at 37 °C until OD₆₀₀ reached ∼0.4. ODs of all cultures were then normalised to OD₆₀₀ of ∼1.0. Cultures were then serially diluted 10⁻¹ to 10⁻⁷ and spotted on LB agar plates containing the relevant antibiotics ± IPTG for induction of each complex combination. Plates were then incubated overnight at 37°C and imaged for colony counting and CFU/ml determination.

Protein expression and purification

For large-scale expression of Salmonella PglZ or co-expression of Salmonella His-SUMO-BrxB and PglZ (Supplementary Fig. S1), E. coli ER2566 was transformed with pSALMZ and pTRB710, respectively. For large-scale expression of E. fergusonii proteins for biochemistry, E. coli ER2566 was transformed with pTRB449 (PglZ) and pTRB444 (BrxB). E. coli Rosetta (DE3) pLysS was also transformed with pTRB444. E. fergusonii mutant derivatives were expressed by transforming E. coli ER2566 with plasmids pTRB729 (PglZ T538A), pTRB730 (PglZ H741A), pTRB763 (PglZ T538A/H741A), pTRB727 (BrxB W135A), and pTRB726 (BrxB R46A), and transforming E. coli Rosetta (DE3) pLysS with pTRB724 (BrxB E47A), pTRB725 (BrxB S133A), pTRB726 (BrxB R46A), pTRB727 (BrxB W135A), and pTRB728 (BrxB E89A).

The same procedures were used for both Salmonella and E. fergusonii proteins. Single colonies were used to inoculate 70 ml 2x YT for overnight growth at 37°C shaking at 180 rpm. Starter cultures were re-seeded 1:100 (v/v) into 1 L 2x YT containing the relevant antibiotic(s) in 2 L baffled flasks and incubated at 37°C until the OD₆₀₀ reached ∼0.4. At this point, the incubation temperature was reduced to 18°C for overnight incubation and expression was induced with IPTG. Cells were harvested by centrifugation at 4200 x g for 20 min at 4°C. Cell pellets were resuspended on ice in ice-cold A500 (20 mM Tris HCl pH 7.9, 500 mM NaCl, 10% (v/v) glycerol, 10 mM imidazole). Resuspended cells were disrupted by sonication (45% amplitude, 10 s on 20 s off pulse intervals, 2 min) and clarified by centrifugation at 45 000 x g for 45 min at 4°C. Clarified cell lysate was loaded onto a 5 ml HisTrap HP column (Cytiva) pre-equilibrated in A100 (20 mM Tris HCl pH 7.9, 100 mM NaCl, 10% (v/v) glycerol, 10 mM imidazole). The HisTrap column was then washed with 50 ml A100, and bound proteins were eluted directly onto a pre-equilibrated 5 ml HiTrap Q HP column using B100 (20 mM Tris HCl pH 7.9, 100 mM NaCl, 10% (v/v) glycerol, 250 mM imidazole). The Q HP column was washed with 50 ml A100 and transferred to an Åkta™ Pure (Cytiva), and the target protein was eluted by anion exchange chromatography using a salt gradient from 100% A100 to 60% C1000 (20 mM Tris HCl pH 7.9, 1 M NaCl, 10% (v/v) glycerol). Chromatographic peak fractions were collected, pooled, and incubated overnight in the presence of human sentrin/SUMO-specific protease 2 (hSENP2) to facilitate the cleavage of the His-SUMO tag at 4°C. The following day, the SENP-treated sample was applied to a second His-Trap HP column pre-equilibrated in A100. The flow-through containing untagged target protein was collected and concentrated by centrifugation using the appropriate MWCO Vivaspin concentrator (Sartorius). Concentrated protein samples were applied to a HiPrep™ 16/60 Sephacryl® S-200 HR column (S-200; Cytiva) pre-equilibrated with 1.2 column volumes (CV) of sizing buffer (50 mM Tris HCl pH 7.9, 500 mM KCl, 10% (v/v) glycerol) for further purification by size exclusion chromatography (SEC). SEC peak fractions were pooled and analysed by SDS-PAGE, then concentrated as described previously and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher). Final purified samples for biochemical analysis were resuspended in a 1:2 mixture of protein sample:storage buffer (50 mM Tris HCl pH 7.9, 500 mM KCl, 70% (v/v) glycerol) and flash frozen in liquid nitrogen for storage at −80°C.

Acinetobacter proteins were expressed as follows: plasmids encoding PglZ^Aci with a C-terminal, thrombin-cleavable Twin-Strep Tag (pET15b.PglZ-TST) and untagged BrxB^Aci (pET24d.BrxB) were used to co-transform E. coli BL21 (DE3) RIL. Overnight cultures (25 ml, LB) were diluted 100-fold into 1 L of LB and grown to an OD₆₀₀ of 0.6, at which time IPTG was added to 200 μM. Cultures were incubated at 16°C for 18 h, pelleted by centrifugation at 4200 x g for 20 min, and the pellets were stored at −20°C. Pellets were lysed by sonication in Buffer W (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA), centrifuged for 25 min at 18 000 x g in an SS34 rotor at 4°C, and the supernatant was filtered through a 5 μm syringe filter. The soluble lysate was bound to streptactin resin, washed with 10 CV of Buffer W and eluted in Buffer E (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin). Samples were concentrated in a 30 kDa MWCO Amicon filter (EMD Millipore) and purified by SEC on a SEC650 column (BioRad) equilibrated in 25 mM Tris-HCl pH 7.5, 200 mM NaCl.

SDS-PAGE electrophoresis

Protein samples were analysed by SDS-PAGE. For protein purity analysis, 4 μg of PglZ and derivative mutants and 4 μg of BrxB and derivative mutants were made up to 10 μl with A100 and mixed with 5 μl 3x sample buffer (187.5 mM Tris HCl pH 6.8, 6% (w/v) SDS, 30% (v/v) glycerol, 0.03% (w/v) bromophenol blue, 150 mM DTT) and denatured for 10 min at 95°C. Protein samples were loaded onto and resolved in 15% (v/v) and 12% (v/v) poly-acrylamide gels, respectively, in 1x Tris-glycine running buffer (25 mM Tris, 192 mM glycine, 0.1% (w/v) SDS) at 180 V for 1h 15 min. For PglZ and BrxB interaction analysis post analytical SEC, 30 μl of fractions for analysis were mixed with 6 μl of 6x sample buffer (375 mM Tris HCl pH 6.8, 12% (w/v) SDS, 60% (v/v) glycerol, 0.06% (w/v) bromophenol blue, 300 mM DTT) and resolved in 15% (v/v) poly-acrylamide gels as described previously. Gels were stained with Quick Coomassie (Protein Ark) and destained with MiliQ. Gel images were obtained on a ChemiDoc™ Imaging System on the Coomassie brilliant blue setting (BioRad).

Protein pull-down assays

His-strep tagged BrxB (expressed from 2HR-T, addgene #29718) was used as bait for pull-down assays of BREX components expressed from pCOLA DUET1 in E. coli BL21 (DE3) pRARE. Overnight cultures were used to inoculate 25 ml of 2xYT with the relevant antibiotics to OD 0.1 before growth at 37°C 180 rpm to OD ∼0.8. Cultures were induced with 1 mM IPTG and incubated at 18°C with shaking overnight. Cells were harvested at 4200 x g for 15 min before freezing at −70 °C. Pellets were defrosted and resuspended in 10 ml 100 mM Tris pH 7.9 150 mM NaCl before sonication (5 min of 10 s pulses at 30% power). Lysates were clarified by centrifugation at 45 000 x g for 10 min at 4°C. Clarified lysates were incubated with 200 μl pre-equilibrated Strep-Tactin Sepharose High Performance resin (Cytiva) at 4°C for 90 min with rolling before application to a Proteus Mini Spin column (ProteinArk). The resin was washed three times with 100 mM Tris pH 7.9 150 mM NaCl, before incubation of the resin in the column with 50 μl 100 mM Tris pH 7.9 150 mM NaCl, 2.5 mM desthiobiotin. The protein was eluted from the column by centrifugation at 12 000 x g for 1 min. The 50 μl eluate was re-applied and re-incubated with the resin before a second centrifugal elution step. Pull-down products were separated and visualised on a 4–15% SDS-PAGE gel.

His-SUMO [28] tagged BrxB was used as bait for pull-down assays following induced expression of all BREX components. Plasmids were co-expressed in T7 E. coli cells for protein complex formation in the following way: empty vector pETDuet1 as control for His-SUMO-BZ (i), empty vectors pETDuet1 and pCDFDuet1 as control for BZ/CX (ii), empty vectors pETDuet1, pCDFDuet1, and pCOLADuet1 as control for BZ/CX/AL (iii) BZ on its own (iv), BZ along with CX (v) and BZ with CX and with AL (vi). Single colonies were inoculated in 20 ml 2x YT for overnight growth at 37°C shaking at 180 rpm. Started cultures were then re-seeded into 1 L 2x YT containing the relevant antibiotic(s) in 2 L baffled flasks and incubated at 37°C until the OD₆₀₀ reached ∼0.4. Cultures were then incubated at 18°C overnight and expression was induced with IPTG. Cells were harvested by centrifugation at 4200 x g for 20 min at 4°C. Cell pellets were resuspended on ice in ice-cold A500. Resuspended cells were disrupted by sonication (45% amplitude, 5 s on 10 s off pulse intervals, 5 min) and centrifuged at 45 000 x g for 30 min at 4°C. Clarified cell lysate was loaded onto a 5 ml HisTrap HP column (Cytiva) pre-equilibrated in A100. The HisTrap column was then washed with 50 ml A100 and transferred to an Åkta™ Pure (Cytiva) for complex elution using an imidazole gradient from 10 mM imidazole to 250 mM using B100. Peak fractions were collected accordingly and concentrated by centrifugation using the MWCO Vivaspin concentrator (Sartorius) of appropriate size. Concentrated complexes were then loaded on a HiPrep™ 16/60 Sephacryl® S-200 HR column (S-200; Cytiva) pre-equilibrated with 1.2 CV of sizing buffer for SEC purification. SEC peak fractions were pooled and analysed by SDS-PAGE and then concentrated and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher).

Bis(4-nitrophenyl) phosphate phosphodiesterase activity assay

EDTA treated PglZ (PglZ EDTA) was prepared by incubating PglZ with 1 mM EDTA for 15 min at room temperature. The EDTA was removed by centrifuging in a 30 kDa MWCO Vivaspin™ ultrafiltration spin column (Cytiva) at 12 000 x g at 4°C, until the volume < 100 μl. The sample was resuspended in ∼400 μl A100 and centrifuged again. This was repeated twice. PglZ and PglZ EDTA (2.2 μM) were incubated with 1x PglZ buffer (“ZB”: 50 mM Tris HCl pH 8.0, 150 mM NaCl) with (PglZ EDTA) or without 0.5 mM MgCl₂, MnCl₂, or CaCl₂, for 30 min at room temperature. The phosphodiesterase reaction was initiated by adding 10 μl 25 mM bis(4-nitrophenyl)phosphate (bis-pNPP, Merck) to 90 μl PglZ reaction mix, and monitoring the release of reaction product, p-nitrophenol, for 2 h at 37°C by measuring the absorbance at 405 nm on a SPECTROstar® Nano microplate reader (BMG Labtech). PglZ derived mutants were also assayed for activity as described. Triplicate reactions were performed per assay, and the assay was completed in triplicate. Control reactions comprised 1x ZB with and without 0.5 mM MgCl₂, MnCl₂, or CaCl₂ in the presence of bis-pNPP.

Nucleotide cleavage assay

PglZ and derivative mutants (2 μM) were incubated with 10 μM ZnCl₂ or MnCl₂ in 1x ZB with 10 μM of the following nucleotides: cyclic hexa-adenosine monophosphate (cA6); cyclic tetra-adenosine monophosphate (cA4); cyclic tri-adenosine monophosphate (cA3); 3′,5′-cyclic di-adenylate (cA2); 5′-phosphoadenylyl-(3′-5′)-adenosine (pApA); 5′-phosphoadenylyl-(3′-5′)-guanosine (pApG); 5′-phosphoguanylyl-(3′-5′)-guanosine (pGpG); 3′,5′-cyclic di-guanylate (cG2); 3′,5′-cyclic adenosine monophosphate (cAMP); 3′,5′-cyclic uridine monophosphate (cUMP); 3′,5′-cyclic thymidine monophosphate (cTMP); 3′,5′-cyclic cytidine monophosphate (cCMP); 3′,5′-cyclic guanosine monophosphate (cGMP); 2′3′-cyclic uridine monophosphate (2′3′ cUMP); 2′3′-cyclic adenosine monophosphate (2′3′ cAMP); 2′3′-cyclic guanosine monophosphate (2′3′ cGMP); cyclic adenosine-(3′-5′)-monophosphate adenosine-(3′-5′)-monophosphate guanosine-(3′-5′)-monophosphate (c[A(3′5′)pA(3′5′)pG(3′5′)p]); cyclic adenosine-(2′-5′)-monophosphate guanosine-(3′-5′)-monophosphate (c[A(2′5′)pG(3′5′)p]); cyclic adenosine-(3′-5′)-monophosphate guanosine-(3′-5′)- monophosphate (c-ApGp); P¹-(5′-adenosyl) P⁴-(5′-adenosyl) tetraphosphate (Ap4A); P¹-(5′-adenosyl) P⁴-(5′adenosyl) triphosphate (Ap3A); or P¹-(5′-adenosyl) P⁴-(5′- guanosyl) tetraphosphate (Ap4G). Reactions were carried overnight at 37°C in a total volume of 50 μl. PglZ (2 μM) was also incubated under the same conditions in the presence of BrxB (10 μM) and BrxB R46A (10 μM). Reactions were centrifuged at 12 000 x g for 10 min at 4°C to remove precipitants and 2 μl was loaded onto an Aeris 5 μm PEPTIDE XB-C18 (150 × 4.6 mm) reversed phase high-performance liquid chromatography (HPLC) column (Phenomenex) at a flow rate of 1.5 ml/min and a linear gradient of 0–30% buffer 2 in 12 column volumes (CV), using buffer 1 (10 mM triethylammonium acetate pH 8.0) and buffer 2 (80% (v/v) acetonitrile, 10 mM triethylammonium acetate pH 8.0) in 12 CV. Protein sample in the absence of nucleotide, and nucleotide in the absence of protein sample were used as controls. Standard mixes contained 10 μM of nucleotide(s) made up to 50 μl in 1x ZB and stored at 4°C.

Inductively coupled plasma mass spectrometry

Total metal contents of protein samples were determined via inductively coupled plasma mass spectrometry (ICP-MS) (Thermo Scientific iCAP RQ ICP-MS) under KED mode (Kinetic Energy Discrimination) utilised with helium. Protein samples were diluted into 2.5% nitric acid containing 10 μg/l berylium, indium and silver as internal standards. Concentrations determined via comparison to matrix-matched elemental standard solutions.

Analytical SEC

Analytical SEC was performed on an Åkta™ Pure FPLC system (Cytiva). Protein samples were made up to 10 μM in a 100 μl final volume with analytical SEC buffer (20 mM Tris HCl pH 7.9, 150 mM NaCl). BrxB was loaded onto a Superdex™ 75 increase 10/300 GL SEC column (S-75i; Cytiva). PglZ and derivative mutants were loaded onto a Superdex™ 200 increase 10/300 GL SEC column (S200i; Cytiva). PglZ and mutant derivatives pre-incubated with BrxB and mutant derivatives at equimolar concentrations for 15 min at room temperature were also loaded onto an S200i. All columns were pre-equilibrated with 1.2 CV analytical SEC buffer. Samples were loaded onto a 100 μl capillary loop using a 100 μl Hamilton syringe. The loop was washed with 500 μl nuclease-free water followed by 500 μl analytical SEC buffer before and between each run using a 500 μl Hamilton syringe. Samples were loaded onto the column by running 500 μl of analytical SEC buffer through the capillary loop at a flow rate of 0.5 ml/min, and samples were resolved on the column using 1.2 CV analytical SEC buffer. In cases where the content of chromatogram peaks required verification by SDS-PAGE or mass spectrometry, 0.5 ml fractionation was performed, and fractions were collected in 96-well deep-plate blocks.

Calibration curves were generated by plotting the elution volumes (V_e) of controls from calibration kits (GE healthcare) against their respective known molecular weights (M_r). Calibration samples were prepared in two individual mixtures, Mix A (3 mg/ml RNase A, Ferritin, Conalbumin, Carbonic Anhydrase) and Mix B (3 mg/ml RNase A, Aldolase, Aprotinin, 4 mg/ml Ovalbumin) and made up to a final volume approximately equal to 0.5% geometric column volume. For determination of column void volume (V_o), 1 mg/ml Blue Dextran was applied to the column as above, with elution volume directly proportional to V_o. Elution volumes (V_e) were calculated using the Peaks function in Unicorn™ 7 (Cytiva) and converted to partitioning coefficients (K_av) using the following equation:

Molecular weight and Stokes radius calibration curves were subsequently plotted using Prism (GraphPad) as K_av versus Log₁₀(M_r,kDa) and Log₁₀(R_st,Å) versus K_av, respectively. Observed R_st values were generated by performing linear regression on respective plots using the following equations:

Observed values were compared against calculated hydrodynamic radii. Radius calculations of inputted AlphaFold3 predictive models were performed using the HullRad tool (Fluidic Analytics).

Mass photometry

Mass Photometry experiments were undertaken on the TwoMP (Refeyn) instrument, using the Acquire 2024R1.1 and Discover 2024R1.0 software for data acquisition and analysis, respectively. DiscoverMP v2024 R2.1 was used to make figures. The autofocus function was used to find the focus plane using 19 μl of phosphate buffered saline (PBS) on uncoated glass slides (Refeyn). Thyroglobulin monomer and dimer peaks, conalbumin and aldolase were used as standards. Stocks of each protein or complex to be tested were prepared at 100 nM immediately before dilution 1:19 into PBS and collection of a 1 min video. Gaussian fits were used for most measurements, with PgZ:BrxB and PglZ:BrxB^E74A measurements also making use of the interval function.

Mass spectrometry

Collected BrxB peaks were buffer exchanged into 10 mM ammonium bicarbonate using a 10 kDa MWCO spin concentrator and submitted for positive ion electrospray time-of-flight mass spectrometry (ES⁺-ToF MS) at a final concentration of 0.5 mg/ml. Analysis was performed at our in-house Durham University Chemistry Department facility by Mr Peter Stokes using a Xevo QToF (Waters, UK) mass spectrometer.

Relevant protein bands were excised from SDS-PAGE gels for their identity to be confirmed via trypsin digest and mass spectrometry by Dr. Adrian Brown at the Department of Biosciences, Durham University.

Pacific biosciences sequencing

Libraries for methylation sequencing were prepared using the SMRTbell HiFi 96 Prep kit (Pacific Biosciences). Bacterial gDNA was sheared using Qiagen Tissue Lyser II at 30 Hz for 240 s to produce DNA fragments with a mean size of 8–10 kb. The DNA was damage and end repaired. SMRT-bell adapters were then ligated. Exonuclease treatment removed non-incorporated SMRT-bell DNA. Sequencing was performed on a PacBio Revio (Pacific Biosciences). Data were analysed using PacBio SMRTAnalysis on SMRTLink_25.1 software Base Modification Analysis for Sequel data, to identify DNA modifications and their corresponding target motifs.

Thermal shift assays

Thermal shift assays (TSAs) were performed to determine the ability of proteins to bind divalent metal cations. Samples of PglZ, PglZ incubated with 1 mM EDTA (PglZ + EDTA), and PglZ treated with EDTA (removed, as described previously; PglZ no metal) were incubated with 4 × 10⁻³ μl SPYRO™ Orange protein dye (ThermoFisher) per 1 μl protein for 1 h at 4°C. Reactions containing 5 μM protein in 1x ZB were incubated with (PglZ no metal) or without 0.5 mM MgCl₂, MnCl₂, CaCl₂, CuCl₂, NiSO₄, and ZnCl₂ for 15 min at RT, made up to 20 μl with nuclease-free water in a sealed 96-well semi-skirted PCR plate (Starlab). Samples were centrifuged and inserted into a CFX connect real-time qPCR machine for thermal shift analysis. The fluorescence was measured in 0.5°C increments from 25°C to 95°C. Deconvolution of thermal shift isotherms was performed using NAMI python tool [29], and thermal shift graphs were generated using Prism (GraphPad).

Nuclease assays

The ability of PglZ to degrade DNA and RNA was analysed using plasmid DNA, phage gDNA, and phage RNA. Prior to any assays, PglZ was treated with EDTA, as described previously, to ensure consistency in the metal binding between samples. For titration experiments, 0, 12, 24, 48, 96, 192, 384, 768, and 1536 nM of purified PglZ were incubated with 6 nM pSG483 supercoiled plasmid DNA, 6 nM pSG483^{BREX KO} supercoiled plasmid DNA (E. fergusonii BREX site mutated), 200 ng pBrxXL WT plasmid DNA, and 200 ng pBrxXL-ΔpglX plasmid DNA. Purified PglZ at 0, 48, 96, 192, 384, 768, and 1536 nM was also incubated with 200 ng φT4 gDNA, 200 ng φPau gDNA, 6 nM M18mp13 ssDNA, and 6 nM φMS2 RNA. Reactions were incubated for 1 h at 37°C with 1x ZB and 0.5 mM MnCl₂. Control reactions either eliminated the metal or included 1 mM ATP in the reaction mix. The activity of PglZ derived mutants against supercoiled pSG483 were tested at 384, 768, and 1536 nM in the presence of 1x ZB and 0.5 mM MnCl₂ at 37°C for 1 h.

To test the activity of PglZ in the presence of various metals, PglZ (768 nM) was incubated with supercoiled pSG483 (6 nM) in 1x ZB at 37°C for 1 h in the presence of 0.5 mM MgCl₂, MnCl₂, CaCl₂, ZnCl₂, CuCl₂, and NiSO₄. Control reactions contained no divalent cations. To test the inhibition of PglZ by various nucleotides, PglZ (768 nM) was incubated supercoiled pSG483 (6 nM) in 1x ZB and 0.5 mM MnCl₂ with and without 1 mM ATP, GTP, CTP, UTP, dATP, dGTP, dTTP, dCTP, ADP, AMP, or AMP-PNP for 1 h at 37°C. Control reactions contained no nucleotide or no PglZ.

To test the activity of PglZ in the presence of BrxB, PglZ (768 nM) was incubated with pSG483 (6 nM) with BrxB at 0.35, 0.7, 1.5, 3.0, and 6.1 μM for 1 h at 37°C. Reactions comprised of 1x ZB and 0.5 mM MnCl₂ and were completed in the presence and absence of 1 mM ATP. PglZ was also incubated with BrxB mutants W135A and R46A. Control reactions contained no protein, PglZ only (768 nM), and BrxB or derivative mutants only (6.1 μM).

All reactions were made up to 20 μl. Reactions were stopped by the addition of 2 μl stopping buffer (5% SDS (v/v), 125 mM EDTA) followed by 4 μl TriTrack loading dye (ThermoFisher). Samples were analysed by agarose gel electrophoresis in 1.4% (w/v) gels for pSG483 analysis, or 0.8% (w/v) gels for pBrxXL, phage gDNA, and RNA analysis using 1x TAE buffer and running at 45 V for ∼16 h. Agarose gels were post stained in 1x TAE containing 0.5 μg/ml ethidium bromide and destained in 1x TAE. Gel images were obtained on a ChemiDoc™ Imaging System on the ethidium bromide setting (BioRad). Gel images were analysed using Fiji (ImageJ; v 2.1.0) with background subtracted. For pSG483 assays, the supercoiled, nicked, and linear DNA band intensity was measured per lane and calculated as a percentage of the total DNA in the respective lane. For pBrxXL assays, the DNA band intensity of all bands per lane were measured independently and compared as a percentage against the corresponding DNA band in the ‘0’ PglZ control lane. For phage gDNA/RNA assays, intact phage gDNA/RNA band intensity was measured in each lane and compared as a percentage against the ‘0’ PglZ control. Mean values and standard deviation were calculated from triplicate data. Data were plotted in Prism (GraphPad).

Efficiency of plating

E. coli bacteriophages were isolated from freshwater sources in Durham city centre and the surrounding areas, as described previously [23]. E. coli DH5α were transformed with pTRB563 (pBrxXL), pTRB564 (pBrxXL-ΔpglX), pTRB744 (pBrxXL-brxB W135A), pTRB745 (pBrxXL-pglZ H741A), pTRB746 (pBrxXL-brxB E47A), pTRB747 (pBrxXL-brxB E89A), pTRB748 (pBrxXL-brxB S133A), pTRB749 (pBrxXL-brxB R46A), pTRB750 (pBrxXL-pglZ T538A), or pTRB766 (pBrxXL-pglZ T538A/H741A) and grown overnight. Serial dilutions of phage Pau were produced in phage buffer (10 mM Tris HCl pH 7.4, 10 mM MgSO₄, 0.01% (v/v) gelatin). 200 μl of overnight culture and 10 μl of phage dilution were added to a sterile 8 ml plastic bijoux with 3 ml of 0.35% (w/v) LB-agar and poured onto LB plates. Plates were incubated overnight at 37°C before plaque forming units (pfu) were counted on each plate. Efficiency of Plating (EOP) values were calculated by determining the phage titre on a test strain divided by the titre on a control strain. EOP data were collected in triplicate and the mean value was plotted in GraphPad Prism.

Modelling the complex of Salmonella PglZ:BrxB via single-particle EM analyses and computational predictions

Negative stain EM

Grids were prepared by applying 4 μl of SEC purified PglZ:BrxB sample at a concentration of approximately 0.04 mg/ml to a glow-discharged Formvar/Carbon 400 mesh Copper grid (Ted Pella). The sample was allowed to absorb for 30 s followed by wicking excess solution with filter paper. The grid was quickly washed two times in 30 μl drops of water and once in a 30 μl drop of 2% uranyl formate (UF) followed by a final staining for 30 s with another 30 μl drop of 2% UF. The grids were air dried for at least 1 h. Grids were screened on an in-house Talos L120C transmission electron microscope (Thermo Fisher), operating at 120 kV and equipped with a 4k x 4k Ceta CMOS high-resolution 16M camera (Thermo Fisher). The sample distributed homogeneously and at random orientations over the surface of the prepared negative stained grids.

Cryo-EM analyses

Flow charts and summary of data collection of the methods described below are shown in Supplementary Fig. S2. Grids were prepared for cryo-EM by applying 3 μl of SEC purified sample at a concentration of 0.25 or 0.5 mg/ml (diluted in 20 mM Tris pH 8.0, 300 mM KCl) to a glow-discharged C-Flat 1.2/1.3 holey carbon film coated copper grid (Electron Microscopy Sciences). The grids were blotted for 5 s at a tension of 0, and plunge-frozen into liquid ethane using a Mark IV Vitrobot (Thermo Fisher). Two datasets of 4686 (dataset 1) and 4731 (dataset 2) movies were collected at a super resolution pixel size of 0.56 Å using a Glacios 200 kV electron microscope (Thermo Fisher) equipped with a Gatan K3 direct electron detector. Preprocessing of datasets was performed in WARP [30] where pixels were binned to 1.122 Å. Datasets were imported into CryoSPARC [31] and particles in dataset 1 were picked by automated searching for Gaussian signals, extracted and Fourier cropped to a box size of 300 and 100 pixels, respectively, and filtered with multiple rounds of 2D classification and selection. Final particles from dataset 1 were lowpass filtered to 20 Å and used as a template for particle picking in dataset 2. Picked particles from dataset 2 were then extracted and filtered as in dataset 1. Final particles from both datasets were combined into a single Ab-initio 3D reconstruction job with 4 classes, resulting in a single class with full particles (124,721) and the remaining classes with fragments or junk particles. The particles contained in the single class were reextracted without Fourier cropping to a box size of 300 pixels followed by homogeneous and non-uniform refinement [32] resulting in a map with Gold Standard Fourier Shell Correlation (GSFSC) resolution of 4.45 Å. The resulting volumes were evaluated in ChimeraX [33].

Despite the reported resolution, and although the overall features of the map were highly consistent with the predicted dimensions of a 1:1 BrxB/PglZ complex as produced by AlphaFold3 (see next section and results), clear features of secondary structure were not visible. Additional examination of the orientation diagnostics for the data set in CryoSPARC produced a conical FSC areas (‘cFAR’) value of 0.04 (Supplementary Fig. S3) and corresponding partical distribution that clearly illustrated severe preferred orientation of the particles, leading to a map that only served to validate the output of computational modelling efforts.

Computational modelling

Predicted models of the Salmonella PglZ:BrxB sub-complex were generated using AlphaFold3 [25] resulting in high per-atom confidences. Initial placement of models into single particle EM density maps was carried out using ChimeraX [33] via the ‘Fit in Map’ tool. Domains were then further fit into the volume individually. The predicted PglZ:BrxB interface was preserved by treating PglZ residues 1-98 as part of the BrxB domain. The remaining residues of PglZ were split into domains for rigid fitting as follow: 99–292, 293–748, and 749–867. The model was then minimally refined with ISOLDE [34] to resolve bond distortions from rigid domain fitting. No rebuilding was performed due to lack of detail in the volumes.

Results

BREX components form larger complexes in vivo

Having previously characterised the structures and function of core Type I BREX components BrxA, PglX, and BrxL [17, 18, 21, 22], we turned to examining interactions between BREX proteins, starting with the system from Salmonella. The BREX locus from Salmonella enterica serovar Typhimurium ST313 strain D23580 (Fig. 1A) had already been sub-cloned and shown to be active in phage defence [17, 23]. Salmonella genes brxA, brxB, brxC, and pglX were cloned into one multiple cloning site of pCOLA DUET1, and genes pglZ and brxL were cloned into the second multiple cloning site. A compatible vector based on 2HR-T (addgene #29718) was generated that expressed His-strep-BrxB. Combining these two vectors, and appropriate vector-only controls, we observed robust expression of the His-strep-BrxB fusion in the absence of the full BREX locus (Fig. 1B). When the His-strep-BrxB fusion was co-expressed with the full Salmonella BREX locus we observed co-purification of BREX proteins BrxC, PglX, PglZ, and BrxL with His-Strep-BrxB, suggesting formation of higher order complexes (Fig. 1B). The indicated bands were confirmed for identity by mass spectrometry (Fig. 1B). The most abundant protein after His-Strep-BrxB was PglZ.

In order to produce larger quantities of the BREX complex(es) the six BREX genes were cloned as three pairs into compatible DUET vectors and co-purification was performed on strains containing increasing combinations of expression vectors (Fig. 1C). For these experiments, the His-strep tag was replaced with a His-SUMO tag to aid later purification. We saw robust His-SUMO-BrxB co-purification with PglZ (lane 2), and then with PglZ, BrxC, and PglX (lane 4), and finally again PglZ, BrxC, PglX, and BrxL (lane 6). No BrxA was co-purified (Fig. 1C). We noted that co-expressing certain combinations of BREX proteins caused poor growth of cells and so performed viable counts (Fig. 1D and Supplementary Fig. S1A). Expression of BrxC and PglX was toxic in E. coli, but this was in part negated by co-expression of His-SUMO-BrxB and PglZ, or BrxA and BrxL, or all six proteins, likely due to formation of larger sub-complexes that mask toxic activities. Due to the robust expression and co-purification of His-SUMO-BrxB and PglZ we chose to pursue this sub-complex for further study. Large scale co-expression and co-purification of His-SUMO-BrxB and PglZ yielded a clean sample of native PglZ:BrxB sub-complexes (Supplementary Figs S1B and C).

Having identified PglZ:BrxB as a strong pairwise protein-protein interaction in Salmonella, we tested this observation further using a previously characterised Type I BREX system found in Acinetobacter [20]. Using Acinetobacter, it was again found that PglZ and BrxB interacted strongly and co-eluted from affinity-based and size exclusion columns (Supplementary Fig. S1D). Interestingly, and unlike its behaviour in Salmonella, BrxB from Acinetobacter was found to strongly require the co-expression and corresponding presence of bound PglZ in order to remain soluble in vitro. Together, these data further imply that the PglZ:BrxB interaction is generalisable and reproducible across Type I BREX systems.

Single particle EM analyses and computational modelling produce a consistent model of the PglZ:BrxB protein complex

SEC of native PglZ, BrxB and co-expressed and co-purified PglZ:BrxB sub-complexes demonstrated an altered elution profile for PglZ:BrxB, indicating formation of a larger sub-complex consistent with the protein pull-down and interaction results described above (Fig. 2A).

Figure 2.

Hybrid model of the Salmonella PglZ:BrxB stable sub-complex. (A) SEC traces of independent Salmonella PglZ and BrxB purifications, and the PglZ:BrxB co-expression sample used for cryo-EM, show that PglZ and BrxB form a stable complex. (B) A cryo-EM map generated from single particles of the Salmonella PglZ:BrxB complex validates the interaction between both proteins, as well as the location of the PglZ:BrxB interaction surfaces and identity of residues in the protein-protein interface (inset box). The resulting model is closely related to the corresponding AlphaFold3 prediction for the Salmonella PglZ:BrxB complex, albeit with rearrangements corresponding to a rotation of the N-terminal domain of PglZ and associated BrxB relative to the larger core of PglZ (Supplementary Fig. S4).

To examine this complex further, we produced a hybrid model of the PglZ:BrxB complex via a combination of single particle EM and computational modelling analyses. The Salmonella PglZ:BrxB sub-complexes were used to perform structural studies through negative stain transmission electron microscopy, followed by cryo-EM (Fig. 2B and Supplementary Fig. S2). The final model displayed a GSFSC resolution of 4.45 Å (Supplementary Fig. S2), but the features of the resulting map, while clearly indicating the presence of a protein volume and dimensions consistent with a PglZ:BrxB complex, did not provide clearly defined features corresponding to individual secondary structural elements, likely caused by the influence of very strong preferred orientation of the particles, with a corresponding cFAR value reported by orientation diagnostics in CryoSPARC of 0.04 (Supplementary Fig. S3).

Despite the limitations of the cryo-EM analyses described above, independently derived models of the PglZ:BrxB complex generated by AlphaFold3 were found to be consistent with the volume and dimensions of the cryo-EM density envelope. BrxB is itself predicted with high confidence to form a globular protein fold, and further predicted to interact with the N-terminal domain (residues 1–96) of PglZ (Fig. 2B). Examination of the model using EMBL PISA [35] identified Salmonella BrxB residues R49, N135 and W137, and PglZ residues K58, E62 and D88, as forming the core of the predicted protein–protein interface (Fig. 2B, inset). The remainder of the PglZ protein is predicted to assume an ‘S’ shape, similar to observed particles from the EM analyses (Supplementary Fig. S2) with a central domain (residues 98–292) and a large C-terminal domain (residues 304–748) that contains the metal-binding site identified within PorX [16], and a final extension including a seven sheet β-barrel (residues 749–867) (Fig. 2B).

The predicted structures of two separate regions from full PglZ:BrxB complexes, produced by five independent runs of AlphaFold3, were each highly consistent across those runs and closely superimposable (Supplementary Fig. S4). Taking the first region (corresponding to the full-length BrxB protein in complex with residues 1 to 96 of PglZ) from the five models and performing a superposition to one another returned an overall RMSD between α-carbons of 0.16 Å (Supplementary Fig. S4A). Similarly, the five models for the second region (corresponding to the remainder of PglZ, extending from residues 99 to 867) also closely superimposed to one another with an RMSD of 0.48 Å (Supplementary Fig. S4B). However, the orientations of those two modelled regions, relative to one another, differ significantly when comparing the same independent AlphaFold3 predictions (Supplementary Fig. S4C), consistent with a possible flexible hinge point in the protein backbone located between residues 96 and 99 of PglZ.

PglZ:BrxB interactions can be ablated by interface mutations

Next, we performed biochemical characterisation of the PglZ:BrxB interface. As experimentation began we noted that the Salmonella PglZ homologue had a tendency to precipitate during tests. Therefore, we chose to use PglZ:BrxB from E. fergusonii, a system we had previously characterised [24], as a substitute biochemical model. Salmonella and E. fergusonii PglZ share 51% amino acid identity, and respective BrxB proteins share 61% amino acid identity.

First, we demonstrated that E. fergusonii PglZ and BrxB also form sub-complexes as previously observed (Figs 1 and 2, Supplementary Fig. S1) for the Salmonella and Acinetobacter homologues. Our hybrid model identified several Salmonella BrxB residues important for the PglZ:BrxB interaction (Fig. 2B). A suite of E. fergusonii BrxB wild type (WT) and mutant proteins were expressed and purified (Supplementary Fig. S5A). E. fergusonii BrxB mutants R46A, S133A and W135A correspond to Salmonella residues R49, N135 and W137, respectively, which were identified as likely important for the PglZ:BrxB interaction within our Salmonella model (Fig. 2B). E. fergusonii mutants E47A and E89A were also made as these residues, though not predicted to be involved in the PglZ:BrxB interaction, are conserved and therefore might inform us of other important roles for BrxB. None of the proteins contained any metals after purification, as analysed by ICP-MS (Supplementary Fig. S5B). SEC analysis of E. fergusonii BrxB WT showed that it formed both monomer and dimer peaks (Supplementary Fig. S5C and D), as confirmed by native mass spectrometric analysis (Supplementary Fig. S5E). Analytical SEC was performed using E. fergusonii BrxB WT, PglZ, and BrxB WT with PglZ (Fig. 3A). Incubating BrxB WT with PglZ caused higher order complexes to form, as shown by elution profiles and corresponding SDS-PAGE analysis of the peaks (Fig. 3A). This indicated that E. fergusonii PglZ:BrxB sub-complexes were also forming, though perhaps with higher order forms being produced beyond those observed for Salmonella and Acinetobacter homologues (Fig. 2A, Supplementary Fig. 1D).

Figure 3.

E. fergusonii BrxB mutants fail to form complexes with PglZ. Analytical SEC (S200i) and SDS-PAGE analysis of PglZ:BrxB complexes formed with BrxB WT (A), BrxB^R46A (B), and BrxB^W135A (C). Samples of PglZ (10 μM) with equimolar BrxB WT and mutants were made up to 100 μl and pre-incubated for 15 min prior to loading on the S200i. The expected elution volumes (V_e) of various complex conformations are highlighted by black or red dotted lines, and the elution profile of PglZ incubated with BrxB is shown as a dark green solid line. Control elution profiles of PglZ alone (light green dashed line) and BrxB alone (dark blue dashed line) are also shown. Fractionated peak samples were resolved on 15% (v/v) polyacrylamide gels for 1 h 15 min in tris-glycine running buffer and stained with Quick Coomassie. Protein identities are highlighted with black arrows. (D) Mass photometry of PglZ with BrxB WT and mutants. Counts were acquired for 60 s with BrxB (5 nM) or samples of PglZ (5 nM) pre-incubated with equimolar BrxB WT and mutants in PBS.

In contrast, co-incubation of E. fergusonii PglZ WT with BrxB R46A and BrxB W135A failed to produce PglZ:BrxB complexes (Fig. 3B and C). BrxB mutants E47A, E89A, and S133A generated intermediate elution profiles, indicating some complexes were forming, but to lesser extent than with BrxB WT (Supplementary Fig. S6). Mass photometric analysis of E. fergusonii PglZ incubated with BrxB WT and mutants showed similar trends, in that complexes formed with BrxB WT, none formed with BrxB R46A, BrxB W135A, or BrxB S133A in these conditions, and complexes formed but less robustly with BrxB E89A and BrxB E47A (Fig. 3D). Collectively, our mutational analysis of the PglZ:BrxB interface supports our hybrid cryo-EM and AlphaFold3 model.

PglZ can cleave cyclic and linear oligonucleotides in a metal-dependent manner

Having examined PglZ:BrxB complex formation, we then wanted to investigate biochemical activity of PglZ first in isolation, then in combination with BrxB. These studies were again performed using the E. fergusonii homologues. A superposition of the AlphaFold3 output for the PglZ domain from E. fergusonii PglZ (residues V474-L759) with the PglZ domain from PorX (PDB: 7PVK, residues T213-K518) produced an RMSD of 3.044 Å (over 1932 atoms) and 2.681 Å (over 178 atoms, C-alpha only) (Fig. 4A). Residues identified as important for PorX metal binding and oligonucleotide cleavage activity, T272 (mutated to T272A in PDB 7PVK) and H500 [16] correspond to E. fergusonii PglZ residues T538 and H741, respectively (Fig. 4A). Mutant proteins E. fergusonii PglZ T538A, H741A and a double mutant T538A/H741A were expressed and purified, and shown to have similar mass photometric and SEC profiles as E. fergusonii PglZ WT (Supplementary Fig. S7).

Figure 4.

PglZ can cleave cyclic nucleotides in a metal-dependent manner. (A) Overlay of AlphaFold3 E. fergusonii PglZ predicted structure with PorX from Porphyromonas gingivalis (PDB: 7PVK). (B) ICP-MS of PglZ WT and mutants showing divalent cations bound following purification. Plotted data represent mean values ± SD. Metal content is plotted as a percentage of the total protein in the sample. (C) TSAs performed upon PglZ WT (5 μM) following EDTA treatment and addition of metals (0.5 mM). Mean changes in melting temperature (ΔTm) are plotted by comparison to PglZ WT in the absence of EDTA or metal, which is set as ‘0’. Black dotted line represents the average melting temperature of the PglZ WT no metal control for comparison with the metal-treated samples. Error bars represent standard deviation (six replicates) (D) Bis(4-nitrophenyl) phosphate (2.5 mM) phosphodiesterase assays using PglZ WT (2 μM) in the presence and absence of EDTA and Mn, and mutants in the presence of Mn. The Mn is supplied by 0.5 mM MnCl₂. Plotted data represent mean values ± SD (nine replicates). Absorbance (A_405nm) represents the amount of reaction product p-nitrophenyl phosphate. (E) HPLC analysis of cyclic hexa-adenosine monophosphate (cA6) cleavage by PglZ WT (2 μM) in the presence of Zn (10 μM). (F) HPLC analysis of cyclic hexa-adenosine monophosphate (cA6) cleavage by PglZ WT (2 μM) treated with EDTA, and mutants T538A and H741A in the presence of Zn (10 μM). Control reactions are represented by cA6 alone, and standard mixes are comprised of pApA and AMP. All nucleotides in the reaction mixes are at 10 μM. Presented traces are representative of triplicate data.

PglZ WT and mutants were tested for metal content following purification, using ICP-MS. There was a clear abundance of zinc in PglZ WT samples, and levels were greatly lowered in the mutant samples (Fig. 4B). To examine the impacts of metals on stability of E. fergusonii PglZ WT and mutants, samples were first stripped of metals by treating the PglZ samples with EDTA (“+EDTA”), and subsequently purified to remove EDTA:metal complexes (‘no metal’) (Fig. 4C and Supplementary Fig. S8). Melting points for stripped samples were compared against those of untreated PglZ WT and mutants (Supplementary Fig. S8E). Having established changes caused by removing metals, the ‘no metal’ samples were then used to examine the impact of providing back a range of metals (Fig. 4C and Supplementary Fig. S8). PglZ WT and mutant proteins were destabilised by copper and, surprisingly, zinc, but were stabilised by magnesium, manganese, and nickel (Fig. 4C and Supplementary Fig. S8). Calcium had no effect, likely because it could not bind (Fig. 4C). Mutants PglZ T538A and PglZ H741A were stabilised by manganese, indicating some metal binding could occur (Supplementary Fig. S8). Double mutant PglZ T538A/H741A was not stabilised by any metal indicating that metal binding in the active site was no longer possible (Supplementary Fig. S8). The double mutant was, however, still destabilised by copper and zinc, suggesting effects for copper and zinc seen with both this mutant and also PglZ WT are due to non-specific binding (Supplementary Fig. S8). The melting temperatures for PglZ WT and mutants indicated that T538A reduces overall stability, but H741A has less of an impact (Supplementary Fig. S8E).

Initial tests for potential phosphodiesterase activity using bis-pNPP as a substrate with PglZ WT and additional zinc resulted in precipitation, and so magnesium, manganese, and calcium were all tested as alternates, with manganese showing the greatest levels of activity (Supplementary Fig. S9). Manganese was therefore selected as an alternate metal in bis-pNPP phosphodiesterase activity assays. Having stripped metals from the samples and restored manganese, E. fergusonii PglZ WT showed robust production of p-nitrophenol from bis-pNPP, at levels greater than for untreated PglZ WT that contained the zinc remaining after purification (Fig. 4D). In contrast, when all mutants were EDTA treated, re-purified, and then provided manganese, PglZ H741A had reduced activity, and both PglZ T538A and the double mutant T538A/H741A lacked activity, demonstrating levels similar to those observed for the PglZ WT sample that was without metal following EDTA treatment (Fig. 4D). This result indicated that PglZ, like PorX, can cleave bis-pNPP in a metal-dependent manner, and that mutations interfering with the metal binding site caused reduced activity.

Cleavage of cyclic oligonucleotides was then tested by incubating E. fergusonii PglZ WT with cA6 and analysing the resulting products by HPLC. Having tested a range of metals, it was noted that zinc was the preferred metal in these assays and was also used at a reduced concentration to prevent protein destabilisation and precipitation. E. fergusonii PglZ WT robustly linearised cA6 and sequentially cleaved nucleotide products, indicated by a trace for each linear species down to AMP (Fig. 4E). PglZ cleavage activity was ablated by EDTA treatment, and mutant PglZ T538A showed no appreciable activity (Fig. 4F). Mutant H741A appeared able to cleave cA6 but did not produce further cleavage products (Fig. 4F). PglZ was then tested against a broader range of nucleotides (Supplementary Figs S10 and S11). PglZ was observed to cleave cyclic oligonucleotides (cA4, cA3, cA2, and cG2) and linear oligonucleotides (pApA, pApG, pGpG, c(ApGp), and c[A(3′5′)pA(3′5′)pG(3′5′)]) containing both adenosine and guanosine, including an oligonucleotide with 2′-5′ rather than 3′-5′ phosphodiester linkages (c[G(2′5′)pA(3′5′)p]). PglZ was unable to cleave cyclic mononucleotides (cAMP, cGMP, cTMP, cUMP, cCMP, 2′3′ cAMP, 2′3′ cGMP, and 2′3′ cUMP) or dinucleotide polyphosphates (Ap3A, Ap4A, and Ap4G) in the presence of Zn. Following this analysis we returned to testing metal usage and noted that at low manganese concentrations we were also able to observe PglZ-dependent cleavage of both cA6 and pApA (Supplementary Fig. S10C). Collectively, these data show robust metal-dependent cyclic and linear oligonucleotide cleavage by PglZ from a BREX system.

PglZ is an endonuclease that can cleave dsDNA

Having established that PglZ has nuclease activity against oligonucleotides we were curious as to whether PglZ could cleave dsDNA. Plasmid pSG483, a pUC19 derivative that can be easily prepared as supercoiled (S), nicked (N), relaxed (R), or linear (L) dsDNA, was selected as a suitable substrate to be tested against E. fergusonii PglZ. Initial assays indicated that manganese would be the preferred metal in this context, but supplementing with zinc did allow some PglZ nuclease activity (Supplementary Fig. S12A). Incubation of pSG483 with a titration of E. fergusonii PglZ WT revealed both nicking and linearisation activities, which were metal-dependent and could be inhibited by ATP (Fig. 5A). This confirmed that PglZ can nick and cut dsDNA, and is an endonuclease. Due to the observed inhibition by ATP, a range of mononucleotides were then tested. All NTPs, dNTPs and AMP-PNP inhibited PglZ nuclease activity, but AMP did not (Supplementary Fig. S12B). The PglZ mutants were then tested for nuclease activity (Fig. 5B). E. fergusonii PglZ H741A had increased nicking but decreased linearisation activity, whereas PglZ T538A and the double mutant PglZ T538A/H741A were both ablated for activity (Fig. 5B). This follows the previous observed trend for activity (Fig. 4F) and indicates T538A prevents metal binding and therefore activity, whilst H741A reduces metal binding and activity.

Figure 5.

PglZ is a metal-dependent nuclease that does not recognise BREX sites. (A) PglZ nicks and linearises supercoiled plasmid pSG483 in a metal-dependent manner that is inhibited by ATP. (B) Mutation of metal binding site alters nuclease activity, either by shifting activity away from linearisation and towards nicking (H741A) or eliminating activity (T538A and T538A/H741A). (C) PglZ can linearise plasmid pSG483 DNA with a mutated BREX site. (D) PglZ does not appear to have site-specific nicking activity on linearised phage DNA, nor be impacted by DNA modifications such as those found in T4. PglZ was titrated against constant supercoiled pSG483 plasmid DNA (6 nM) or phage gDNA (200 ng) in the presence and absence of MnCl₂ (0.5 mM) and ATP (1 mM). Samples were incubated for 1 h. Control lanes represent supercoiled (S), nicked (N), linear (L), and relaxed (multiple topoisomers; R) plasmid DNA, or contain the appropriate DNA in the absence of protein. Assays are presented on 1.4% (w/v) or 0.8% (w/v) agarose 1x TAE gels post-stained with ethidium bromide. Assays shown are representative of triplicate experiments. Data points and error bars represent the mean ± SD of triplicate data.

The BREX methyltransferase, PglX, determines sequence recognition for host methylation and phage defence [17]. E. fergusonii PglX recognises the sequence GCTAAT, and there is 1 copy of this motif in pSG483. A mutant pSG483 was generated (pSG483^{BREX KO}) with the GCTAAT motif mutated to GCTATT to allow testing of whether PglZ cleavage is dependent on BREX motifs. When E. fergusonii PglZ WT was titrated against pSG483^{BREX KO}, there was no observed difference to the result with pSG483 (Fig. 5A and C). We then considered whether BREX methylation might impact PglZ nuclease activity. We prepared pBREXxl WT, a plasmid encoding the full E. fergusonii locus and the mutant pBREXxl-ΔpglX. Each plasmid has previously been shown to be BREX methylated and lacking methylation, respectively [24]. E. fergusonii PglZ WT caused equal degradation of both plasmids, indicating BREX methylation does not impact PglZ activity under these isolated conditions (Supplementary Fig. S12C and D).

Phage Pau was shown to be susceptible to E. fergusonii BREX in an earlier study [24]. Phage Pau does not have modified DNA [23], unlike phage T4, which has modified cytosines and so is inherently resistant to BREX. When tested, E. fergusonii PglZ was able to cleave genomic DNAs from both these phages (Fig. 5D). The cleavage did not produce a distinct pattern, rather a faint smear of products, demonstrating that PglZ is likely a sequence-independent endonuclease (Fig. 5D). It was also interesting that T4 cytosine modifications did not impact PglZ cleavage when in isolation, though T4 (and other modified phages) are resistant to BREX phage defence. E. fergusonii PglZ WT could also cause sequence-independent cleavage of ssDNA, using M18mp13 genomic DNA as substrate (Supplementary Fig. S13A). There was no activity, however, on MS2 phage genomic RNA (Supplementary Fig. S13B).

Finally, to ensure our biochemical data and structural study are aligned, we confirmed that Salmonella PglZ WT also demonstrated metal-dependent nicking and linearisation of pSG483 (Supplementary Fig. S13C). Salmonella PglZ showed a preference for zinc or magnesium and in contrast to E. fergusonii PglZ, Salmonella PglZ could also use calcium and copper, and could not use manganese (Supplementary Fig. S13C).

Interaction with BrxB impacts neither PglZ nuclease activity nor inhibition of nuclease activity by ATP

Gel-based nuclease activities were then used to investigate the impact of BrxB interactions on PglZ activity. In these assays, E. fergusonii BrxB had no identifiable nicking or linearisation activity (Fig. 6A). Titration of E. fergusonii BrxB against E. fergusonii PglZ had no appreciable impact on PglZ nicking and linearisation activities until the highest BrxB concentration (Fig. 6A). Comparisons of the AlphaFold3 model for the BrxB structure using DALI [36] indicated some similarity to nucleotide binding regions of AAA + proteins, but BrxB appears to be lacking key Walker motif residues. As BrxB had the potential for binding ATP, we investigated whether BrxB might impact the observed inhibition of PglZ activity by ATP (Fig. 5A). When the same PglZ to BrxB titration was performed in the presence of ATP there was no indication that BrxB could overcome inhibition of PglZ activity by ATP (Fig. 6A). For completeness, we also tested whether non-interacting E. fergusonii BrxB mutants impacted PglZ activity, but no effect was observed (Fig. 6B). Next, we used HPLC analysis of oligonucleotide cleavage as another measure of BrxB impact. Neither BrxB WT nor BrxB R46A altered the ability of E. fergusonii PglZ to cleave cA6 or pApA (Fig. 6C). This indicates that the role of BrxB, at least in these isolated conditions, is independent of PglZ nuclease activity, and likely has more relevance in the context of larger BREX complexes.

Figure 6.

BrxB interacting with PglZ does not appreciably alter PglZ nuclease activity. (A) Incubation of PglZ (768 nM) with a titration of BrxB WT in the absence or presence of MnCl₂ (0.5 mM) and ATP (1 mM). (B) Incubation of PglZ WT (768 nM) with a titration of BrxB mutants R46A and W135A in the absence or presence of MnCl₂ (0.5 mM). Samples were incubated for 1 h. Control lanes represent supercoiled (S), nicked (N), linear (L), and relaxed (multiple topoisomers; R) plasmid DNA, or contain DNA in the absence of protein. Assays are presented on 1.4% (w/v) agarose 1x TAE gels post-stained with ethidium bromide. Assays shown are representative of triplicate experiments. (C) PglZ (2 μM) incubated in the presence of BrxB (10 μM) does not prevent cleavage of cA6 or pApA (10 μM). Control reactions are comprised of the nucleotide in the absence of protein. Standard mixes are comprised of pApA and AMP. Presented traces are representative of triplicate data.

We also examined by analytical SEC whether any of the E. fergusonii PglZ mutants T538A, H741A, or T538A/H741A would impact BrxB WT binding and formation of higher order complexes. As expected, due to these mutations being distant from the BrxB binding site (Figs 2B and 4A), no impact on complex formation was observed (Supplementary Fig. S14).

PglZ nuclease activity contributes to BREX phage defence but not BREX-dependent methylation

With the E. fergusonii PglZ and BrxB mutations now characterised biochemically we examined their impact on the two measurable BREX phenotypes, phage defence and BREX-dependent methylation. Mutations were constructed in the context of pBrxXL, which encodes the full E. fergusonii BREX locus under native promoters [24]. The suite of mutants were tested for defence against phage Pau from the Durham collection [23], measured by EOP, using an appropriate vector control. The positive and negative controls pBrxXL and pBrxXL-ΔpglX provided strong and no phage defence, respectively (Fig. 7A). BrxB mutant constructs S133A and W135A, and the PglZ H741A construct all showed a small reduction in phage defence activity, around 10-fold (Fig. 7A). PglZ T538A and double mutant T538A/H741A constructs showed a large reduction in defence of around 3 logs, but remained impressively active (Fig. 7A). These data indicate that mutations preventing PglZ:BrxB interactions or ablating PglZ nuclease activity have an impact but can be compensated for in vivo.

Figure 7.

PglZ:BrxB mutants have small impact on BREX phage defence and no impact on BREX methylation. (A) EOP results of phage Pau tested against E. fergusonii BREX constructs. Error bars represent standard deviation from the mean of triplicate data. (B) PacBio sequencing results showing the percentage of BREX motif methylation in each strain.

Genomic DNA was extracted from each strain and PacBio sequencing analysis was performed to investigate BREX-dependent methylation, examining N6mA methylation on the fifth adenine of the GCTAAT motif (Fig. 7B). Positive control pBrxXL WT had 99.4% methylation of BREX motifs, and negative control pBrxXL-ΔpglX had no detectable methylation, as observed previously [24]. In all cases, the current mutations in pBrxXL had no impact on BREX-dependent methylation (Fig. 7B), indicating PglZ nuclease activity works independently from BREX methylation.

Discussion

Type I BREX systems perform phage defence and BREX-dependent methylation, but the mechanisms for each activity have proven difficult to uncover. Having previously performed individual studies of BREX components and regulators [17–22], in this study, we examined what higher order BREX complexes form in cells and focussed on characterisation of PglZ, BrxB and the resulting stable PglZ:BrxB sub-complex.

Using Salmonella BrxB as bait we observed pull-down of BrxB with BrxC, PglX, PglZ, and also BrxL (Fig. 1). Although BrxL is the effector protein needed for phage defence with the E. coli and Acinetobacter BREX systems [20, 37], we previously noted that Salmonella BREX can provide phage defence without BrxL [23], and there is potential for BrxL having a regulatory role in this system [38]. It was therefore curious that BrxL readily associates with other Salmonella BREX components. However, low levels of BrxL can also be seen in similar pull-down experiments performed using E. coli BREX [18]. Using BrxL as bait does also pull-down other BREX components, though the data do show that the more robust complex appears to be formed of BrxB, BrxC, PglX, and PglZ [18]. We noted strong association between PglZ and BrxB, also seen with E. coli BREX [18], and so chose to examine this sub-complex as a step towards understanding the structure and function of higher order BREX complexes.

Our resulting hybrid model of Salmonella PglZ:BrxB (Fig. 2) combines electron density obtained by cryo-EM with models generated by AlphaFold3 [25] of the two proteins in complex with one another. The model suggests a point of structural flexibility that allow movement of the PglZ:BrxB interaction domain (Supplementary Fig. S4).

The role of BrxB has thus far remained hypothetical, yet the predicted fold mimics AAA + nucleotide binding domains. Having examined the structure of Salmonella PglZ:BrxB, we switched to E. fergusonii PglZ and BrxB for biochemical characterisation as the proteins behaved more reproducibly under assay conditions. We were able to demonstrate binding of E. fergusonii BrxB to PglZ, and pinpoint residues required for stable complex formation (Fig. 3 and Supplementary Fig. S6). When we later tested whether BrxB might bind nucleotides and alter the observed inhibition of PglZ by ATP, nothing changed (Fig. 6). Alternatively, and due to the observed data in Salmonella (Fig. 1) and E. coli [18], we postulate that the role of BrxB might be as a scaffold protein, participating within and allowing connections between multiple BREX components within higher order complexes.

The first evidence of biochemical activity for PglZ domains, originally considered an alkaline phosphatase [15], came from demonstration that the PglZ domain of PorX, a two-component signalling system response regulator, could act as a phosphodiesterase and linearise cyclic nucleotides [16]. PorX activity is zinc-dependent. After substantial efforts to identify a preferred metal and concentration, it could be demonstrated that E. fergusonii PglZ has similar activity to PorX, cleaving a wide range of cyclic nucleotides, but not cyclic mononucleotides or dinucleotide polyphosphates (Fig. 4 and Supplementary Fig. S7–11).

We hypothesised that our observed E. fergusonii PglZ activity could impact dsDNA. When tested, E. fergusonii PglZ nicked and linearised dsDNA (Fig. 5). This activity was applicable to ssDNA, but not dsRNA, and was independent of BREX motifs, appeared sequence-independent, and was not impacted by BREX methylation or by larger DNA modifications such as glucose modifications to hydroxymethylated cytosines in phage T4 genomic DNA (Fig. 5 and Supplementary Figs S12 and S13). Initial investigations of BREX activity indicated little digestion of invading phage DNA, merely an inhibition of phage genome replication, making BREX a classic ‘restriction’ system [15]. Nicking is a hallmark of multiple other phage defence systems, including Shedu, Lamassu, Dnd, and Gabija [39–43]. In the latter case, Gabija activity is also regulated by the detection and degradation of nucleotides [43]. As we observed inhibition of PglZ nicking activity in the presence of ATP (Fig. 5), we cannot rule out analogous regulatory activity by PglZ in the context of a full BREX mechanism. It is unclear whether ATP might be competing for the PglZ domain catalytic site, or have another inhibitory binding site that alters activity. We also cannot dismiss a potential role for PglZ-dependent nicking in protecting from invading DNA, perhaps as a precursor licensing step to allow further inhibition of replication.

When assaying the two phenotypes of phage defence and methylation we saw that ablation of PglZ:BrxB interactions made only a small impact on phage defence (Fig. 7A), perhaps because within a higher order complex other interactions occur to support function. In contrast, removal of PglZ nuclease activity by mutation had a stronger impact though still did not remove all phage defence activity, indicating that the overall BREX mechanism can compensate for loss or reduction in PglZ function (Fig. 7A). Whilst deletion of pglZ prevents BREX phage defence and methylation [17, 20, 37], our mutations of either PglZ nuclease activity or BrxB binding had no impact on methylation (Fig. 7B), indicating PglZ likely plays a role in formation of the BREX methylation complex, but that its nuclease activity is not also required for target methylation.

A working model for BREX activity was recently posited, wherein a BREX-BCXZ complex would form and move along DNA to allow methylation [18]. Our data support formation of this complex (Fig. 1), and contrary to data indicating PglX alone is sufficient for methylation [44], we could not see methylation using Salmonella PglX nor the same E. coli homologue [17, 18], and expression of PglX alone in cells also does not result in methylation [37]. A BREX-BCXZ complex would have to be able to distinguish between DNA templates containing BREX methylation on one strand following host genome replication, and target invading DNA containing no BREX methylation. When invading DNA is recognised by PglX, we envision subsequent activation of the PglZ nuclease. The resulting nicking of target DNA would then license the BREX-BCXZ complex to switch from methylation surveillance to restriction. This could include recruitment of BrxL and, for instance, movement of the BREX complex to stall replication forks. Nevertheless, there remains many questions as to the specifics of BREX activity and our data indicate clear next steps towards the characterisation of higher order BREX complexes.

Supplementary data

Supplementary data are available at NAR online.

Conflict of interest

T.R.B. is an employee of, and B.L.S. is a paid consultant for, New England Biolabs, which provided funding support for this study and which develops a wide variety of phage restriction systems for commercial sale.

Funding

This work was supported by a Biotechnology and Biological Sciences Research Council Newcastle-Liverpool-Durham Doctoral Training Partnership studentship [grant number BB/T008695/1] to J.J.R., a Biotechnology and Biological Sciences Research Council responsive mode grant [grant number BB/Y003659/1] to M.P., a Lister Institute Prize Fellowship to A.K. and T.R.B., New England Biolabs (NEB), the Fred Hutchinson Cancer Center (FHCC) and the NIH for both BLS (R01GM105691 and R35GM148166) and BKK (R15GM140375).

For the purpose of open access, the authors have applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission. Funding to pay the Open Access publication charges for this article was provided by BBSRC Newcastle, Liverpool, Durham Doctoral Training Partnership.

Data availability

The cryo-EM model and corresponding maps for PglZ complexed with BrxB have been deposited in the RCSB PDB database (ID code 9NV3) and in the EMDB (ID code EMD-49827). All other data needed to evaluate the conclusions in the paper are present in the paper and/or Supplementary Data.