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

Phage defence, Bacteriophage exclusion, nuclease, metal-dependent, BREX 17 18 19 20 21 22 23 24 25 26 27 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 2

28 BREX (Bacteriophage Exclusion) systems, identified through shared identity with Pgl (Phage Growth 29 Limitation) systems, are a widespread, highly diverse group of phage defen ce systems found 30 throughout bacteria and archaea. The varied BREX Types harbour multiple protein subunits (between 31 four and eight) and all encode a conserved putative phosphatase (PglZ aka BrxZ) and an equally 32 conserved, putative ATPase (BrxC). Almost all BREX systems also contain a site -specific 33 methyltransferase (PglX aka BrxX). Despite having determined the structure and fundamental 34 biophysical and biochemical behavio urs for the PglX methyltransferase, the BrxL effector, the BrxA 35 DNA-binding protein and the BrxR transcriptional regulator, the mechanism by which BREX impedes 36 phage replication remains largely undetermined. In this study, we identify a stable BREX sub-complex 37 of PglZ:BrxB, validate the structure and dynamic behavio ur of that sub-complex, and assess the 38 biochemical activity of PglZ, revealing it to be a metal -dependent nuclease. PglZ can cleave cyclic 39 oligonucleotides, linear oligonucleotides, plasmid DNA and both non -modified and modified linear 40 phage genomes. PglZ nuclease activity has no obvious role in BREX-dependent methylation, but does 41 contribute to BREX phage defence. BrxB binding does not impact PglZ nuclease activity. These data 42 contribute to our growing understanding of the BREX phage defence mechanism. 43 44 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 3

45 Up to 10% of bacterial and archaeal genes are dedicated to phage defen ce (1). The mechanisms 46 employed to defend against phage infection are diverse and include prevention of viral entry, 47 induction of cell dormancy or death upon infection, and mechanisms that degrade viral genomes or 48 block viral DNA replication (2, 3). Phages combat these systems through the evolution of elaborate 49 countermeasures that block their action, leading to viral resistance and a continuous arms race 50 between phage and bacterial population s (4). Recent analyses have demonstrated that bacteria 51 encode far more phage defence systems than just the most well-studied ‘first responder’ systems such 52 as restriction endonucleases and CRISPR (5–10). Furthermore, many of these newly discovered 53 bacterial systems display obvious similarities to human innate viral defense systems, implying 54 common evolutionary origins and related mechanisms of action (11). 55 56 Originally discovered in the early 1980s (12), Phage Growth Limitation (Pgl) (13, 14) and 57 related Bacteriophage Exclusion (BREX) systems are widespread in bacterial and archaeal species (15). 58 BREX systems are encoded by single operons, often within genetic defence islands, and are currently 59 categorised into at least six types based on the number of genes in each system (typically four to eight) 60 and on the sequence -based functional annotation of those individual genes and putative translated 61 protein subunits (15); Type I systems, the most common subtype, comprise six conserved genes and 62 can readily be assayed for the two phenotypes of phage defence and BREX -dependent methylation 63 though the mechanisms are unknown. 64 65 Pgl and BREX systems have two genes in common. The first is named pglZ (aka brxZ) and the 66 second is brxC (15). The PglZ domain of a two-component signalling system response regulator, PorX, 67 has recently been shown to degrade cyclic nucleotides (16), but any equivalent activity within BREX 68 has not yet been explored. Beyond PglZ and BrxC, most BREX systems include a gene encoding a site-69 specific methyltransferase, termed PglX (aka BrxX). The identity and order of the remaining genes in 70 each BREX type vary significantly: various BREX systems encode protein subunits with domains that 71 display recognisable homology to kinases, phosphatases, DNA and/or nucleotide binding domains, 72 DNA modification enzymes, chambered AAA+ ATPases, and/or DNA helicases. Several BREX subunits 73 are quite large, with significant regions of unknown structure -function properties and behavio urs 74 flanking domains with well -annotated putative functions. Type I BREX systems, like their related 75 counterparts, do not contain any readily identifiable DNA nuclease domains or subunits and appear 76 to likely restrict phage by inhibiting phage DNA replication within the infected bacterial cell. 77 78 Despite having previously determined the high resolution structures and biochemical 79 activities of the PglX methyltransferase (17, 18), BrxR (a WYL -domain helix-turn-helix DNA binding 80 transcriptional regulator) (19, 20), BrxA (a small DNA binding protein) (21), and BrxL (a chambered 81 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 4 AAA+ ATPase and dsDNA binding protein) (22), the mechanism by which BREX systems function to 82 restrict phage replication and to protect the host genome from the system’s activit y is still largely 83 unknown. We have performed analyses to further characterise several Type I BREX systems we have 84 previously investigated: those from Salmonella Typhimurium (17, 23) and Escherichia fergusonii (24) 85 (Fig. 1A), and from Acinetobacter (20, 22). Using these systems, in vivo pull-down and co-expression 86 analysis identified larger BREX compl exes and a stable sub -complex formed by PglZ and BrxB. 87 Computational models of the PglZ:BrxB interactions and their likely conformation and dynamic 88 behaviour when bound to one another was validated through single -particle cryoEM analy ses. 89 Subsequent biochemical analysis has identified that PglZ recapitulates PorX activity and is a metal -90 dependent nuclease that can cleave not only a broad range of cyclic and linear oligonucleotides, but 91 also plasmid and linear dsDNA. The BrxB interaction does not impact PglZ nuclease activity, nor is a 92 nuclease required for BREX -dependent methylation. Nuclease activity does, however, contribute to 93 BREX phage defence. These data contribute to our growing understanding of the elusive BREX 94 mechanism. 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 5 112 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 6

113 Bacterial strains and culture conditions 114 E. coli strains DH5α (Invitrogen), ER2796 (New England Biolabs) (25), ER2566 (New England Biolabs), 115 Rosetta 2 (DE3) pLysS (Novagen) , BL21 (DE3) pRARE (Novagen) , BL21 (DE3) RIL (Novagen) and T7 116 Express (New England Biolabs) were routinely grown at 37 C, either on agar plates or shaking at 150 117 rpm for liquid cultures. 2x Yeast Extract Tryptone (YT) was used as the standard growth media for 118 liquid cultures, and Luria Broth (LB) was supplemented with 0.35% (w/v) or 1.5% (w/v) agar for semi-119 solid and solid agar plates, respectively. When necessary, growth media was supplemented with 120 ampicillin (Ap, 100 μg/ml), chloramphenicol (Cm, 25 μg/ml), kanamycin ( Km, 100 μg/mL, 121 spectinomycin ( Sp, 100 μg/mL), isopropyl--D-thiogalactopyranoside (IPTG, 1 mM). Growth was 122 monitored using a spectrophotometer (WPA Biowave C08000) measuring optical density at 600 nm 123 (OD600). 124 125 DNA isolation and manipulation 126 Plasmid DNA was purified from transformed DH5 cells using an NEB Monarch® Plasmid MiniPrep kit 127 following the manufacturer’s instructions. Larger amounts of negatively supercoiled plasmid pSG483 128 (26) DNA for assays was purified from transformed DH5  cells using a Machery -Nagel NucleoBond 129 Xtra Midi Plus EF kit following the manufacturer’s instructions. Plasmid DNA was eluted in MiliQ and 130 stored at -20 C. Plasmids are described in Supplementary Table S1. 131 132 Phage genomic DNA was purified by incubating 450 μl phage lysate with 4.5 μl DNase I (1 133 mg/ml; Sigma-Aldrich) and 12.5 μl RNase A (10 mg/ml; ThermoFisher) for 30 min at 37 C. The lysate 134 was further incubated with 2.25 μl proteinase K (20 mg/ml; Sigma-Aldrich) and 23 μl of 10% (w/v) SDS 135 for 30 min at 37 C. The sample was mixed with 500 μl UltraPure phenol:chloroform:isoamyl alcohol 136 (25:24:1; v/v/v) (ThermoFisher) and centrifuged at 16,000 x g for 5 min at 4 C. The aqueous layer was 137 removed carried forward, and the previous step was repeated. The resulting aqueous layer was mixed 138 with 500 μl chloroform:isoamyl alcohol (24:1; v/v) and centrifuged at 16,000 x g for 5 min at 4 C. The 139 aqueous layer was carried forward and incubated with 45 μl 3 M sodium acetate pH 5.2 and 500 μl 140 isopropanol for 15 min at room temperature, before being centrifuged at 16,000 x g for 20 min and 4 141 C. The supernatant was removed, and the pellet was washed with 70% ethanol by gentle aspiration 142 before being dried at room temperature. The dry pellet was soaked in 50 μl of MiliQ and incubated 143 overnight at 4 C. The gDNA was analysed on a 0.75% 1x TAE agarose gel by agarose gel 144 electrophoresis, and stored at -20 C. 145 146 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 7 Preparation of nicked, linear, and relaxed form pSG483 147 Linear pSG483 was obtained through incubation of 10 μg pSG483 with 10 units of BamHI -HF® (New 148 England Biolabs) in 1x CutSmart buffer (New England Biolabs) for 1 h at 37 C. The enzyme was 149 deactivated by incubation at 65 C for 10 min. Nicked pSG483 was obtained by incubating 10 -50 μg 150 pSG483 with 10 units Nb.Bpu10I (ThermoFisher) in 1x Buffer R ( ThermoFisher) for 4 h at 37 C. The 151 reaction was terminated by incubation at 80 C for 20 min. 152 153 For production of relaxed pSG483, 50 μg nicked pSG483 was further incubated with 1 mM ATP 154 and 10 units T4 ligase ( New England Biolabs) for 16 h at room temperature. After ligation, an equal 155 volume of UltraPure phenol:chloroform:isoamyl alcohol (25:24:1; v/v/v) (ThermoFisher) was added 156 to the reaction mixture, vortexed briefly, and centrifuged at 16 ,000 x g for 2 min. The resulting 157 aqueous layer was removed and carried forward. An equal volume of chloroform was added to the 158 aqueous layer before centrifugation at 16,000 x g for 2 min. The resulting aqueous layer was carried 159 forward and 1/10 volume of 3 M sodium acetate pH 5.2 was added, followed by 2 volumes of 100% 160 ethanol. The sample was mixed by pipetting and stored at -80 C for 30 min. The sample was 161 centrifuged at 16,000 x g for 20 min at 4 C. The ethanol was removed, and the DNA pellet dried at 162 room temperature. The DNA pellet was resuspended to 300 ng/μl with MiliQ. All DNA products were 163 analysed by agarose gel electrophoresis prior to storage at -20 C. 164 165 Bacterial growth assays 166 T7 Express E. coli cells were transformed with: empty vectors pETDuet, pCDFDuet and pCOLADuet (i), 167 pTRB710 (His-SUMO-BrxB and PglZ (BZ)) (ii), pTRB759 (BrxC and PglX (CX)) (iii), pTRB758 (BrxA and 168 BrxL (AL)) (iv) and combinations of BZ/CX (v), BZ/AL (vi), CX/AL (vii) and BZ/CX/AL (viii). Colonies were 169 inoculated and grown overnight in 5 ml 2x YT with respective antibiotics at 37 C shaking at 180 rpm. 170 Cultures were re-seeded 1:100 (v/v) in 100 ml 2x YT with the relevant antibiotics and grown at 37 C 171 until OD600 reached ~0.4. ODs of all cultures were then normalised to OD600 of ~1.0. Cultures were then 172 serially diluted 10-1 to 10-7 and spotted on LB agar plates containing the relevant antibiotics +/ - IPTG 173 for induction of each complex combination. Plates were then incubated overnight at 37 C and imaged 174 for colony counting and CFU/ml determination. 175 176 Protein expression and purification 177 For large-scale expression of Salmonella PglZ or co-expression of Salmonella His-SUMO-BrxB and PglZ, 178 E. coli ER2566 was transformed with pSALMZ and pTRB710, respectively. For large-scale expression of 179 E. fergusonii proteins for biochemistry, E. coli ER2566 was transformed with pTRB449 (PglZ) and 180 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 8 pTRB444 (BrxB). E. coli Rosetta (DE3) pLysS was also transformed with pTRB444. E. fergusonii mutant 181 derivatives were expressed by transforming E. coli ER2566 with plasmids pTRB729 ( PglZ T538A), 182 pTRB730 (PglZ H741A), pTRB763 ( PglZ T538A/H741A), pTRB727 ( BrxB W135A), and pTRB726 ( BrxB 183 R46A), and transforming E. coli Rosetta (DE3) pLysS with pTRB724 (BrxB E47A), pTRB725 (BrxB S133A), 184 pTRB726 (BrxB R46A), pTRB727 (BrxB W135A), and pTRB728 (BrxB E89A). 185 186 The same procedures were used for both Salmonella and E. fergusonii proteins. Single colonies 187 were used to inoculate 70 ml 2x YT for overnight growth at 37 C shaking at 180 rpm. Starter cultures 188 were re-seeded 1:100 (v/v) into 1 L 2x YT containing the relevant antibiotic(s) in 2 L baffled flasks and 189 incubated at 37 C until the OD600 reached ~0.4. At this point, the incubation temperature was reduced 190 to 18 C for overnight incubation and expression was induced with IPTG. Cells were harvested by 191 centrifugation at 4,200 x g for 20 min at 4 C. Cell pellets were resuspended on ice in ice -cold A500 192 (20 mM Tris HCl pH 7.9, 500 mM NaCl, 10% (v/v) glycerol, 10 mM imidazole). Resuspended cells were 193 disrupted by sonication (45% amplitude, 10 s on 20 s off pulse intervals, 2 min) and clarified by 194 centrifugation at 45,000 x g for 45 min at 4 C. Clarified cell lysate was loaded onto a 5 ml HisTrap HP 195 column (Cytiva) pre-equilibrated in A100 (20 mM Tris HCl pH 7.9, 100 mM NaCl, 10% (v/v) glycerol, 10 196 mM imidazole). The HisTrap column was then washed with 50 ml A100, and bound proteins were 197 eluted directly onto a pre-equilibrated 5 ml HiTrap Q HP column using B100 (20 mM Tris HCl pH 7.9, 198 100 mM NaCl, 10% (v/v) glycerol, 250 mM imidazole). The Q HP column was washed with 50 ml A100 199 and transferred to an Åkta Pure (Cytiva), and the target protein was eluted by anion exchange 200 chromatography (AEC) using a salt gradient from 100% A100 to 60% C1000 (20 mM Tris HCl pH 7.9, 1 201 M NaCl, 10% (v/v) glycerol). Chromatographic peak fractions were collected, pooled, and incubated 202 overnight in the presence of human sentrin/SUMO -specific protease 2 (hSENP2) to facilitate the 203 cleavage of the His-SUMO tag at 4 C. The following day, the SENP -treated sample was applied to a 204 second His-Trap HP column pre -equilibrated in A100. The flow -through containing untagged target 205 protein was collected and concentrated by centrifugation using the appropriate MWCO Vivaspin 206 concentrator (Sartorius). Concentrated protein samples were applied to a HiPrep 16/60 Sephacryl® 207 S-200 HR column (S -200; Cytiva) pre -equilibrated with 1.2 column volumes (CV) of sizing buffer (50 208 mM Tris HCl pH 7.9, 500 mM KCl, 10% (v/v) glycerol) for further purification by size exclusion 209 chromatography (SEC). SEC peak fractions were pooled and analysed by SDS-PAGE, then concentrated 210 as described previously and quantified using a NanoDrop 2000 Spectrophotometer (Thermo Fisher). 211 Final purified samples for biochemical analysis were resuspended in a 1:2 mixture of protein 212 sample:storage buffer (50 mM Tris HCl pH 7.9, 500 mM KCl, 70% (v/v) glycerol) and flash frozen in 213 liquid nitrogen for storage at -80 C. 214 215 Acinetobacter proteins were expressed as follows: plasmids encoding PglZAci with a C-terminal, 216 thrombin-cleavable Twin-Strep Tag (pET15b.PglZ-TST) and untagged BrxBAci (pET24d.BrxB) were used 217 to co-transform E. coli BL21 (DE3) RIL. Overnight cultures (25 ml, LB) were diluted 100-fold into 1 L of 218 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 9 LB and grown to an OD 600 of 0.6, at which time IPTG was added to 200 µM. Cultures were incubated 219 at 16 °C for 18 h, pelleted by centrifugation at 4,200 x g for 45 min, and the pellets were stored at -20 220 °C. Pellets were lysed in Buffer W (100 mM Tris -HCl pH 8.0, 150 mM NaCl , 1 mM EDTA), centrifuged 221 for 25 min at 18,000 x g in an SS34 rotor at 4 °C, and the supernatant was filtered through a 5 µm 222 syringe filter. The soluble lysate was bound to streptactin resin, washed with 10 CV of Buffer W and 223 eluted in Buffer E (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin). Samples 224 were concentrated in a 30 kDa MWCO Amicon filter (EMD Millipore) and purified by SEC on a SEC650 225 column (BioRad) equilibrated in 25 mM Tris-HCl pH 7.5, 200 mM NaCl. 226 227 SDS-PAGE electrophoresis 228 Protein samples were analysed by SDS -PAGE. For protein purity analysis, 4 μg of PglZ and derivative 229 mutants, and 4 μg of BrxB and derivative mutants were made up to 10 μl with A100 and mixed with 5 230 μl 3x sample buffer (187.5 mM Tris HCl pH 6.8, 6% (w/v) SDS, 30% (v/v) glycerol, 0.03% (w/v) 231 bromophenol blue, 150 mM DTT) and denatured for 10 min at 95 C. Protein samples were loaded 232 onto and resolved in 15% (v/v) and 12% (v/v) poly -acrylamide gels, respectively, in 1x Tris-glycine 233 running buffer (25 mM Tris, 192 mM glycine, 0.1% (w/v) SDS) at 180 V for 1h 15 min. For PglZ and BrxB 234 interaction analysis post analytical SEC, 30 μl of fractions for analysis were mixed with 6 μl of 6x sample 235 buffer (375 mM Tris HCl pH 6.8, 12% (w/v) SDS, 60% (v/v) glycerol, 0.06% (w/v) bromophenol blue, 236 300 mM DTT) and resolved in 15% (v/v) poly-acrylamide gels as described previously. Gels were 237 stained with Quick Coomassie (Protein Ark) and destained with MiliQ. Gel images were obtained on a 238 ChemiDoc Imaging System on the Coomassie brilliant blue setting (BioRad). 239 240 Protein pull-down assays 241 His-strep tagged BrxB (expressed from 2HR-T, addgene #29718) was used as bait for pull-down assays 242 of BREX components expressed from pCOLA DUET1 in E. coli BL21 (DE3) pRARE. Overnight cultures 243 were used to inoculate 25 ml of 2xYT with the relevant antibiotics to OD 0.1 before growth at 37 oC 244 180 rpm to OD ~0.8. Cultures were induced with 1 mM IPTG and incubated at 18 oC with shaking 245 overnight. Cells were harvested at 4,200 x g for 15 min before freezing at -70 oC. Pellets were defrosted 246 and resuspended in 10 ml 100 mM Tris pH 7.9 150 mM NaCl before sonication (5 min of 10 s pulses at 247 30% power). Lysates were clarified by centrifugation at 45,000 x g for 10 min at 4 C. Clarified lysates 248 were incubated with 200 µl pre-equilibrated Strep-Tactin Sepharose High Performance resin (Cytiva) 249 at 4 oC for 90 min with rolling before application to a Proteus Mini Spin column (ProteinArk). The resin 250 was washed three times with 100 mM Tris pH 7.9 150 mM NaCl, before incubation of the resin in the 251 column with 50 µl 100 mM Tris pH 7.9 150 mM NaCl, 2.5 mM desthiobiotin. The protein was eluted 252 from the column by centrifugation at 12,000 x g for 1 min . The 50 µl eluate was re -applied and re -253 incubated with the resin before a second centrifugal elution step. Pull-down products were separated 254 and visualised on a 4- 15% SDS-PAGE gel. 255 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 10 256 His-SUMO (27) tagged BrxB was used as bait for pull -down assays following induced 257 expression of all BREX components. Plasmids were co-expressed in T7 E. coli cells for protein complex 258 formation in the following way: empty vector pETDuet1 as control for His-SUMO-BZ (i), empty vectors 259 pETDuet1 and pCDFDuet1 as control for BZ/CX (ii), empty vectors pETDuet1, pCDFDuet1 and 260 pCOLADuet1 as control for BZ/CX/AL (iii) BZ on its own (iv), BZ along with CX (v) and BZ with CX and 261 with AL (vi). Single colonies were inoculated in 20 m l 2x YT for overnight growth at 37 C shaking at 262 180 rpm. Started cultures were then re-seeded into 1 L 2x YT containing the relevant antibiotic(s) in 2 263 L baffled flasks and incubated at 37 C until the OD600 reached ~0.4. Cultures were then incubated at 264 18 C overnight and expression was induced with IPTG. Cells were harvested by centrifugation at 4,200 265 x g for 20 min at 4 C. Cell pellets were resuspended on ice in ice-cold A500. Resuspended cells were 266 disrupted by sonication (45% amplitude, 5 s on 10 s off pulse intervals, 5 min) and centrifuged at 267 45,000 x g for 30 min at 4 C. Clarified cell lysate was loaded onto a 5 m l HisTrap HP column (Cytiva) 268 pre-equilibrated in A100. The HisTrap column was then washed with 50 m l A100, and transferred to 269 an Åkta Pure (Cytiva) for complex elution using an imidazole gradient from 10 mM imidazole to 250 270 mM using B100. Peak fractions were collected accordingly and concentrated by centrifugation using 271 the MWCO Vivaspin concentrator (Sartorius) of appropriate size. Concentrated complexes were then 272 loaded on a HiPrep 16/60 Sephacryl® S-200 HR column (S-200; Cytiva) pre-equilibrated with 1.2 CV 273 of sizing buffer for size exclusion chromatography (SEC) purification. SEC peak fractions were pooled 274 and analysed by SDS -PAGE, then concentrated and quantified using a NanoDrop 2000 275 Spectrophotometer (Thermo Fisher). 276 277 Bis(4-nitrophenyl) phosphate phosphodiesterase activity assay 278 EDTA treated PglZ (PglZ EDTA) was prepared by incubating PglZ with 1 mM EDTA for 15 min at room 279 temperature. The EDTA was removed by centrifuging in a 30 kDa MWCO Vivaspin ultrafiltration spin 280 column (Cytiva) at 12,000 x g at 4 C, until the volume < 100 μl. The sample was resuspended in ~400 281 μl A100 and centrifuged again. This was repeated twice. PglZ and PglZ EDTA (2.2 μM) were incubated 282 with 1x PglZ buffer (“ZB”: 50 mM Tris HCl pH 8.0, 150 mM NaCl) with (PglZ EDTA) or without 0.5 mM 283 MgCl2, MnCl 2, or CaCl 2, for 30 mins at room temperature. The phosphodiesterase reaction was 284 initiated by adding 10 μl 25 mM bis(4-nitrophenyl)phosphate (bis-pNPP, Merck) to 90 μl PglZ reaction 285 mix, and monitoring the release of reaction product, p-nitrophenol, for 2 hours at 37 C by measuring 286 the absorbance at 405 nm on a SPECTROstar® Nano microplate reader (BMG Labtech). PglZ derived 287 mutants were also assayed for activity as described. Triplicate reactions were performed per assay, 288 and the assay was completed in triplicate. Control reactions comprised 1x ZB with and without 0.5 289 mM MgCl2, MnCl2, or CaCl2 in the presence of bis-pNPP. 290 291 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 11 Nucleotide cleavage assay 292 PglZ and derivative mutants (2 μM) were incubated with 10 μM ZnCl2 or MnCl2 in 1x ZB with 10 μM of 293 the following nucleotides: cyclic hexa -adenosine monophosphate (cA6); cyclic tetra -adenosine 294 monophosphate (cA4); cyclic tri-adenosine monophosphate (cA3); 3 ,5-cyclic di-adenylate (cA2); 5-295 phosphoadenylyl-(3-5)-adenosine (pApA); 5 -phosphoadenylyl-(3-5)-guanosine (pApG); 5 -296 phosphoguanylyl-(3-5)-guanosine (pGpG); 3 ,5-cyclic di -guanylate (cG2); 3 ,5-cyclic adenosine 297 monophosphate (cAMP); 3 ,5-cyclic uridine monophosphate (cUMP); 3 ,5-cyclic thymidine 298 monophosphate (cTMP); 3 ,5-cyclic cytidine monophosphate (cCMP); 3 ,5-cyclic guanosine 299 monophosphate (cGMP); 2 3-cyclic uridine monophosphate (2 3 cUMP); 2 3-cyclic adenosine 300 monophosphate (23 cAMP); 23-cyclic guanosine monophosphate (23 cGMP); cyclic adenosine-(3-301 5)-monophosphate adenosine -(3-5)-monophosphate guanosine -(3-5)-monophosphate 302 (c[A(35)pA(35)pG(35)p]); cyclic adenosine -(2-5)-monophosphate guanosine -(3-5)-303 monophosphate (c[A(2 5)pG(35)p]); cyclic adenosine -(3-5)-monophosphate guanosine -(3-5)- 304 monophosphate (c-ApGp); P1-(5-adenosyl) P4-(5-adenosyl) tetraphosphate (Ap4A); P 1-(5-adenosyl) 305 P4-(5adenosyl) triphosphate (Ap3A); or P 1-(5-adenosyl) P 4-(5- guanosyl) tetraphosphate (Ap4G). 306 Reactions were carried overnight at 37 C in a total volume of 50 μl. PglZ (2 μM) was also incubated 307 under the same conditions in the presence of BrxB (10 μM) and BrxB R46A (10 μM). Reactions were 308 centrifuged at 12,000 x g for 10 min at 4 C to remove precipitants and 2 μl was loaded onto an Aeris 309 5 m PEPTIDE XB-C18 (150 x 4.6 mm) reversed phase high-performance liquid chromatography (HPLC) 310 column (Phenomenex) at a flow rate of 1.5 ml/min and a linear gradient of 0-30% buffer 2 in 12 column 311 volumes (CV), using buffer 1 (10 mM triethylammonium acetate pH 8.0) and buffer 2 (80% (v/v) 312 acetonitrile, 10 mM triethylammonium acetate pH 8.0) in 12 CV. P rotein sample in the absence of 313 nucleotide, and nucleotide in the absence of protein sample were used as controls. Standard mixes 314 contained 10 μM of nucleotide(s) made up to 50 μl in 1x ZB and stored at 4 C. 315 316 Inductively coupled plasma mass spectrometry (ICP-MS) 317 Total metal contents of protein samples were determined via ICP-MS (Thermo Scientific iCAP RQ ICP-318 MS) under KED mode (Kinetic Energy Discrimination) utilised with helium. Protein samples were 319 diluted into 2.5% nitric acid containing 10 μg/l berylium, indium and silver as internal standards. 320 Concentrations determined via comparison to matrix-matched elemental standard solutions. 321 322 Analytical size exclusion chromatography 323 Analytical size exclusion chromatography (SEC) was performed on an Åkta Pure FPLC system 324 (Cytiva). Protein samples were made up to 10 μM in a 100 μl final volume with analytical SEC buffer 325 (20 mM Tris HCl pH 7.9, 150 mM NaCl). BrxB was loaded onto a Superdex 75 increase 10/300 GL SEC 326 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 12 column (S -75i; Cytiva). PglZ and derivative mutants were loaded onto a Superdex  200 increase 327 10/300 GL SEC column (S200i; Cytiva). PglZ and mutant derivatives pre -incubated with BrxB and 328 mutant derivatives at equimolar concentrations for 15 min at room temperature were also loaded 329 onto an S200i. All columns were pre -equilibrated with 1.2 CV analytical SEC buffer. Samples were 330 loaded onto a 100 μl capillary loop using a 100 μl Hamilton syringe. The loop was washed with 500 μl 331 nuclease-free water followed by 500 μl analytical SEC buffer before and between each run using a 500 332 μl Hamilton syringe. Samples were loaded onto the column by running 500 μl of analytical SEC buffer 333 through the capillary loop at a flow rate of 0.5 ml/min, and samples were resolved on the column 334 using 1.2 CV analytical SEC buffer. In cases where the content of chromatogram peaks required 335 verification by SDS -PAGE or mass spectrometry, 0.5 ml fractionation was performed, and fractions 336 were collected in 96-well deep-plate blocks. 337 338 Calibration curves were generated by plotting the elution volumes ( Ve) of controls from 339 calibration kits (GE healthcare) against their respective known molecular weights ( Mr). Calibration 340 samples were prepared in 2 individual mixtures, Mix A (3 mg /ml RNase A, Ferritin, Conalbumin, 341 Carbonic Anhydrase) and Mix B (3 mg/ml RNase A, Aldolase, Aprotinin, 4 mg/ml Ovalbumin) and made 342 up to a final volume approximately equal to 0.5% geometric column volume. For determination of 343 column void volume ( Vo), 1 mg /ml Blue Dextran was applied to the column as above, with elution 344 volume directly proportional to Vo. Elution volumes (Ve) were calculated using the Peaks function in 345 Unicorn 7 (Cytiva) and converted to partitioning coefficients (Kav) using the following equation: 346 𝐾𝑎𝑣 = 𝑉𝑒 − 𝑉𝑜 𝑉𝑐 − 𝑉𝑜 347 Molecular weight and Stokes radius calibration curves were subsequently plotted using Prism 348 (GraphPad) as Kav vs Log 10(Mr,kDa) and Log 10(Rst,Å) vs Kav, respectively. Observed Rst values were 349 generated by performing linear regression on respective plots using the following equations: 350 𝑀𝑟 = 10∧ (𝐾𝑎𝑣 − 𝑐 𝑚 ) 351 352 𝑅𝑠𝑡 = 10∧ ((𝑚(𝐾𝑎𝑣) + 𝐶) 353 Observed values were compared against calculated hydrodynamic radii. Radius calculations 354 of inputted AlphaFold predictive models were performed using the HullRad tool (Fluidic Analytics). 355 356 Mass photometry 357 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 13 Mass Photometry experiments were undertaken on the TwoMP ( Refeyn) instrument, using the 358 Acquire 2024R1.1 and Discover 2024R1.0 software for data acquisition and analysis , respectively. 359 DiscoverMP v2024 R2.1 was used to make figures. The autofocus function was used to find the focus 360 plane using 19 µl of PBS on uncoated glass slides (Refeyn). Thyroglobulin monomer and dimer peaks, 361 conalbumin and aldolase were used as standards. Stocks of each protein or complex to be tested were 362 prepared at 100 nM immediately before dilution 1:19 into PBS and collection of a 1 min video. 363 Gaussian fits were used for most measurements, with PgZ:BrxB and PglZ:BrxBE74A measurements also 364 making use of the interval function. 365 366 Mass spectrometry 367 Collected BrxB peaks were buffer exchanged into 10 mM ammonium bicarbonate using a 10 kDa 368 MWCO spin concentrator and submitted for positive ion electrospray time -of-flight mass 369 spectrometry (ES+-ToF MS) at a final concentration of 0.5 mg /ml. Analysis was performed at our in-370 house Durham University Chemistry Department facility by Mr Peter Stokes using a Xevo QToF 371 (Waters, UK) mass spectrometer. 372 373 Relevant protein bands were excised from SDS -PAGE gels for their identity to be confirmed 374 via trypsin digest and mass spectrometry by Dr. Adrian Brown at the Department of Biosciences, 375 Durham University. 376 377 Pacific biosciences sequencing 378 Libraries for methylation sequencing were prepared using the SMRTbell HiFi 96 Prep kit (Pacific 379 Biosciences). Bacterial gDNA was sheared using Qiagen Tissue Lyser II at 30 Hz for 240 s to produce 380 DNA fragments with a mean size of 8 –10 kb. The DNA was damage and end repaired. SMRT -bell 381 adapters were then ligated. Exonuclease treatment removed non-incorporated SMRT -bell DNA. 382 Sequencing was performed on a PacBio Revio (Pacific Biosciences). Data were analysed using PacBio 383 SMRTAnalysis on SMRTLink_25.1 software Base Modification Analysis for Sequel data, to identify DNA 384 modifications and their corresponding target motifs. 385 386 Thermal shift assays (TSAs) 387 Thermal shift assays (TSAs )were performed to determine the ability of proteins to bind divalent 388 metal cations. Samples of PglZ, PglZ incubated with 1 mM EDTA (PglZ + EDTA), and PglZ treated with 389 EDTA (removed, as described previously; PglZ - EDTA) were incubated with 4 x 10-3 μl SPYRO 390 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 14 Orange protein dye (ThermoFisher) per 1 μl protein for 1 h at 4 C. Reactions containing 5 μM 391 protein in 1x ZB were incubated with (PglZ - EDTA) or without 0.5 mM MgCl2, MnCl2, CaCl2, CuCl2, 392 NiSO4, and ZnCl2 for 15 mins at RT, made up to 20 μl with nuclease-free water in a sealed 96-well 393 semi-skirted PCR plate (Starlab). Samples were centrifuged and inserted into a CFX connect real-time 394 qPCR machine for thermal shift analysis. The fluorescence was measured in 0.5 C increments from 395 25 to 95 C. Deconvolution of thermal shift isotherms was performed using NAMI python tool (28), 396 and thermal shift graphs were generated using Prism (GraphPad). 397 398 Nuclease assays 399 The ability of PglZ to degrade DNA and RNA was analysed using plasmid DNA, phage gDNA, and phage 400 RNA. Prior to any assays, PglZ was treated with EDTA, as described previously, to ensure consistency 401 in the metal binding between samples. For titration experiments, 0, 12, 24, 48, 96, 192, 384, 768, and 402 1536 nM of purified PglZ were incubated with 6 nM pSG483 supercoiled plasmid DNA, 6 nM 403 pSG483BREX KO supercoiled plasmid DNA (E. fergusonii BREX site mutated), 200 ng pBrxXL WT plasmid 404 DNA, and 200 ng pBrxXL -pglX plasmid DNA. Purified PglZ at 0, 48, 96, 192, 384, 768, and 1536 nM 405 was also incubated with 200 ng T4 gDNA, 200 ng Pau gDNA, 6 nM M18mp13 ssDNA, and 6 nM 406 MS2 RNA. Reactions were incubated for 1 h at 37 C with 1x ZB and 0.5 mM MnCl2. Control reactions 407 either eliminated the metal or included 1 mM ATP in the reaction mix. The activity of PglZ derived 408 mutants against supercoiled pSG483 were tested at 384, 768, and 1536 nM in the presence of 1x ZB 409 and 0.5 mM MnCl2 at 37 C for 1 h. 410 411 To test the activity of PglZ in the presence of various metals, PglZ (768 nM) was incubated with 412 supercoiled pSG483 (6 nM) in 1x ZB at 37 oC for 1 h in the presence of 0.5 mM MgCl 2, MnCl2, CaCl2, 413 ZnCl2, CuCl2, and NiSO4. Control reactions contained no divalent cations. To test the inhibition of PglZ 414 by various nucleotides, PglZ (768 nM) was incubated supercoiled pSG483 (6 nM) in 1x ZB and 0.5 mM 415 MnCl2 with and without 1 mM ATP, GTP, CTP, UTP, dATP, dGTP, dTTP, dCTP, ADP, AMP, or AMP -PNP 416 for 1 h at 37 C. Control reactions contained no nucleotide or no PglZ. 417 418 To test the activity of PglZ in the presence of BrxB, PglZ (768 nM) was incubated with pSG483 419 (6 nM) with BrxB at 0.35, 0.7, 1.5, 3.0, and 6.1 μM for 1 h at 37 oC. Reactions comprised of 1x ZB and 420 0.5 mM MnCl2 and were completed in the presence and absence of 1 mM ATP. PglZ was also incubated 421 with BrxB mutants W135A and R46A. Control reactions contained no protein, PglZ only (768 nM), and 422 BrxB or derivative mutants only (6.1 μM). 423 424 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 15 All reactions were made up to 20 μl. Reactions were stopped by the addition of 2 μl stopping 425 buffer (5% SDS (v/v), 125 mM EDTA) followed by 4 μl TriTrack loading dye (ThermoFisher). Samples 426 were analysed by agarose gel electrophoresis in 1.4% (w/v) gels for pSG483 analysis, or 0.8% (w/v) 427 gels for pBrxXL, phage gDNA, and RNA analysis using 1x TAE buffer and running at 45 V for ~16 h. 428 Agarose gels were post stained in 1x TAE containing 0.5 μg/ml ethidium bromide and destained in 1x 429 TAE. Gel images were obtained on a ChemiDoc  Imaging System on the ethidium bromide setting 430 (BioRad). Gel images were analysed using Fiji (ImageJ; v 2.1.0) with background subtracted. For 431 pSG483 assays, the supercoiled, nicked, and linear DNA band intensity was measured per lane and 432 calculated as a percentage of the total DNA in the respective lane. For pBrxXL assays, the DNA band 433 intensity of all bands per lane were measured independently and compared as a percentage against 434 the corresponding DNA band in the ‘0’ PglZ control lane. For phage gDNA/RNA assays, intact phage 435 gDNA/RNA band intensity was measured in each lane and compared as a percentage against the ‘0’ 436 PglZ control. Mean values and standard deviation were calculated from triplicate data. Data were 437 plotted in Prism (GraphPad). 438 439 Efficiency of plating (EOP) 440 E. coli bacteriophages were isolated from freshwater sources in Durham city centre and the 441 surrounding areas, as described previously (23). E. coli DH5 were transformed with pTRB563 442 (pBrxXL), pTRB564 (pBrxXL -pglX), pTRB744 (pBrxXL -brxB W135A), pTRB745 (pBrxXL-pglZ H741A), 443 pTRB746 (pBrxXL-brxB E47A), pTRB747 (pBrxXL-brxB E89A), pTRB748 (pBrxXL-brxB S133A), pTRB749 444 (pBrxXL-brxB R46A), pTRB750 ( pBrxXL-pglZ T538A), or pTRB766 (pBrxXL -pglZ T538A/H741A) and 445 grown overnight. Serial dilutions of phage Pau were produced in phage buffer (10 mM Tris HCl pH 7.4, 446 10 mM MgSO4, 0.01% (v/v) gelatin). 200 μl of overnight culture and 10 μl of phage dilution were added 447 to a sterile 8 ml plastic bijoux with 3 ml of 0.35% (w/v) LB-agar and poured onto LB plates. Plates were 448 incubated overnight at 37 C before plaque forming units (pfu) were counted on each plate. EOP 449 values were calculated by determining the phage titre on a test strain divided by the titre on a control 450 strain. EOP data were collected in triplicate and the mean value was plotted in GraphPad Prism. 451 452 Initial single particle screening of Salmonella PglZ:BrxB 453 Negative stain grids were prepared by applying 4 μl of size exclusion chromatography (SEC) purified 454 PglZ:BrxB sample at a concentration of approximately 0.04 mg /ml to a glow -discharged 455 Formvar/Carbon 400 mesh Copper grid (Ted Pella). The sample was allowed to absorb for 30 s 456 followed by wicking excess solution with filter paper. The grid was quickly washed two times in 30 μl 457 drops of water and once in a 30 μl drop of 2% uranyl formate (UF) followed by a final staining for 30 s 458 with another 30 μl drop of 2% UF. The grids were air dried for at least 1 hr. Grids were screened on an 459 in-house Talos L120C transmission electron microscope (Thermo Fisher), operating at 120 kV and 460 equipped with a 4k x 4k Ceta CMOS high -resolution 16M camera (Thermo Fisher). The sample 461 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 16 distributed homogeneously and at random orientations over the surface of the prepared negative 462 stained grids. 463 464 CryoEM hybrid model determination for Salmonella PglZ:BrxB 465 Flow charts and summary of data collection of the methods described below are shown in 466 Supplementary Figure S2. Grids were prepared for cryoEM by applying 3 μl of SEC purified sample at 467 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 468 C-Flat 1.2/1.3 holey carbon film coated copper grid (Electron Microscopy Sciences). The grids were 469 blotted for 5 s at a tension of 0, and plunge-frozen into liquid ethane using a Mark IV Vitrobot (Thermo 470 Fisher). Two datasets of 4686 (dataset 1) and 4731 (dataset 2) movies were collected at a super 471 resolution pixel size of 0.56 Å using a Glacios 200 kV electron microscope (Thermo Fisher) equipped 472 with a Gatan K3 direct electron detector. Preprocessing of datasets was performed in WARP (29) 473 where pixels were binned to 1.122 Å. Datasets were imported into CryoSPARC (30) and particles in 474 dataset 1 were picked by automated searching for Gaussian signals, extracted and Fourier cropped to 475 a box size of 300 and 100 pixels, respectively, and filtered with multiple rounds of 2D classification and 476 selection. Final particles from dataset 1 were lowpass filtered to 20 Å and used as a template for 477 particle picking in dataset 2. Picked particles from dataset 2 were then extracted and filtered as in 478 dataset 1. Final particles from both datasets were combined into a single Ab-initio 3D reconstruction 479 job with 4 classes, resulting in a single class with full particles (124,721) and the remaining classes with 480 fragments or junk particles. The particles contained in the single class were reextracted without 481 Fourier cropping to a box size of 300 pixels followed by homogeneous and non -uniform refinement 482 (31) resulting in a map with GSFSC resolution of 4.45 Å. The resulting volumes were evaluated in 483 ChimeraX (32). 484 485 Model fitting 486 Predicted models of the Salmonella PglZ:BrxB sub-complex was generated by AlphaFold (33) resulting 487 in high per-atom confidences. Initial placement of models was accomplished in ChimeraX (32) using 488 the Fit in Map tool. Domains were then further fit into the volume individually. The predicted PglZ:BrxB 489 interface was preserved by treating PglZ residues 1 -98 as part of the BrxB domain. The models were 490 then further refined in the Phenix suite (34) using Cryo_fit (35) and Real Space Refine (36). No 491 rebuilding was performed due to lack of detail in the volumes. 492 493 494 495 496 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 17

497 Salmonella BREX components form larger complexes in vivo 498 Having previously performed characterisation of independent core Type I BREX components BrxA, 499 PglX and BrxL (17, 18, 21, 22), we turned to examining interactions between BREX proteins. The BREX 500 locus from Salmonella Typhimurium strain D23580 (Fig. 1A) had already been sub-cloned and shown 501 to be active in phage defence (17, 23). Salmonella genes brxA, brxB, brxC and pglX were cloned into 502 one multiple cloning site of pCOLA DUET1, and genes pglZ and brxL were cloned into the second 503 multiple cloning site. A compatible vector based on 2HR -T (addgene #29718) was generated that 504 expressed His -strep-BrxB. Combining these two vectors, and appropriate vector -only controls, we 505 observed robust expression of the His-strep-BrxB fusion in the absence of the full BREX locus (Fig. 1B). 506 When then expressed in cells also expressing the full Salmonella BREX locus we observed co -507 purification of BREX proteins BrxC, PglX, PglZ and BrxL with His -Strep-BrxB, indicating formation of 508 higher order complexes (Fig. 1B ). The indicated bands were confirmed for identity by mass 509 spectrometry (Fig. 1B). The most abundant protein after His-Strep-BrxB was PglZ. In order to produce 510 larger quantities of the BREX complex(es) the six BREX genes were cloned as three pairs into 511 compatible DUET vectors and co -purification was performed on strains containing increasing 512 combinations of expression vectors ( Fig. 1C). For these experiments, the His -strep tag was replaced 513 with a His-SUMO tag to aid later purification. We saw robust His-SUMO-BrxB co-purification with PglZ, 514 and then with PglZ, BrxC and PglX, and finally again PglZ, BrxC, PglX and BrxL. No BrxA was co-purified 515 (Fig. 1C). We noted that certain combinations of vector caused poor growth of cells and so performed 516 viable counts (Fig. 1D and Supplementary Fig. S1A). Expression of BrxC and PglX was toxic in E. coli, 517 but this was in part negated by co-expression of His-SUMO-BrxB and PglZ, or BrxA and BrxL, or all six 518 proteins. Due to the robust expression and co -purification of His -SUMO-BrxB and PglZ we chose to 519 pursue this sub-complex for further study. Large scale co-expression and co-purificaiton of His-SUMO-520 BrxB and PglZ yielded a clean sample of native PglZ:BrxB sub-complexes (Supplementary Figs. S1B-C). 521 522 Having identified PglZ:BrxB as a strong pairwise protein-protein interaction in Salmonella, we 523 further tested this observation using a previously characteri sed Type I BREX system found in 524 Acinetobacter (20). Using that system, it was also found that PglZ and BrxB interact strongly and co -525 eluted from affinity -based and size exclusion columns in a 1:1 ratio ( Supplementary Fig. S1 ). 526 Interestingly, and unlike its behavio ur in Salmonella, BrxB from Acinetobacter was found to require 527 the co-expression and corresponding presence of bound PglZ in order to remain soluble in vitro. These 528 data indicate that the PglZ:BrxB interaction is generalisable and reproducible. 529 530 Salmonella PglZ:BrxB complexes show dynamic movement 531 Size exclusion chromatography (SEC) of native PglZ, BrxB and co-expressed and co-purified PglZ:BrxB 532 expected sub-complexes demonstrated an altered elution profile for PglZ:BrxB, indicating formation 533 of a larger sub-complex (Fig. 2A). The Salmonella PglZ:BrxB sub-complexes were used to perform 534 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 18 535 536 537 538 539 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 19 structural studies through negative stain transmission electron microscopy, followed by cryo-electron 540 microscopy (cryoEM) (Fig. 2B and Supplementary Fig. S2). The final model had a Gold Standard Fourier 541 Shell Correlation ( GSFSC) resolution of 4.45 Å ( Supplementary Fig. S2 ). At this resolution it was 542 possible to make use of AlphaFold outputs for PglZ:BrxB to generate a final hybrid model of the 543 PglZ:BrxB sub-complex. BrxB is itself a globular protein and was shown to be bound to the N-terminal 544 domain (residues 1-96) of PglZ (Fig. 2B). EMBL PISA (37) identified BrxB residues R49, N135 and W137, 545 and PglZ residues K58, E62 and D88, as important for binding (Fig. 2B, inset). PglZ forms an “S” shape, 546 with a central domain (residues 98-292) and a large C -terminal PglZ domain (residues 30 4-748) that 547 contains the metal -binding site identified within PorX (16), and a final extension including a seven 548 sheet β-barrel (residues 749-867) (Fig. 2B). Comparison of the Salmonella PglZ:BrxB AlphaFold model 549 alone against our cryoEM hybrid model shows two distinct points of movement (Supplementary Fig. 550 S3). The PglZ N -terminal domain and BrxB, and the C -terminal β-barrel extension have both made 551 large movements between the two models, whereas the central and PglZ domains remain fixed 552 (Supplementary Fig. S3). This dynamic flexibility is likely the cause of our data being limited to lower 553 resolution. Nevertheless, this model confirms the presence of a flexible but stable PglZ:BrxB complex. 554 555 PglZ can cleave cyclic and linear oligonucleotides in a metal -dependent 556 manner 557 Next, we performed biochemical characterisation of PglZ in isolation, in preparation for later 558 investigation of the PglZ:BrxB sub-complex. As experimentation began we noted that the Salmonella 559 PglZ homologue had a tendency to precipitate during tests. As such, we chose to use PglZ from E. 560 fergusonii, a system we had previously characterised (24), as a substitute biochemical model. 561 562 A superposition of the AlphaFold output for the PglZ domain from E. fergusonii PglZ (residues 563 V474-L759) with the PglZ domain from PorX (PDB: 7PVK, residues 213 -518) produced an RMSD of 564 2.822 (over 1096 atoms) ( Figs. 2B and 3A). Residues identified as important for PorX metal binding 565 and oligonucleotide cleavage activity, T272 (mutated to T272A in PDB 7PVK) and H500 (16) correspond 566 to E. fergusonii PglZ residues T538 and H741, respectively (Fig. 3A). Mutant proteins E. fergusonii PglZ 567 T538A, H741A and a double mutant T538A/H741A were expressed and purified, and shown to have 568 similar mass photometric and SEC profiles as E. fergusonii PglZ WT (Supplementary Fig. S4). 569 570 PglZ wild type (WT) and mutants were tested for metal content following purification, using 571 inductively coupled plasma mass spectrometry (ICP-MS). There was a clear abundance of zinc in PglZ 572 WT samples, and levels were greatly lowered in the mutant samples ( Fig. 3B ). Following EDTA 573 treatment to remove metals and subsequent purification to remove EDTA, E. fergusonii PglZ WT and 574 mutants were tested for stability in the presence of a range of metals using thermal shift assays (TSAs) 575 (Fig. 3C and Supplementary Fig. S5). PglZ proteins were destabilised by copper and, surprisingly, zinc, 576 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 20 577 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 21 but were stabilised by magnesium, manganese and nickel (Fig. 3C and Supplementary Fig . S5). 578 Calcium had no effect, likely because it could not bind ( Fig. 3C). Mutants PglZ T538A and PglZ H741A 579 were stabilised by manganese, indicating some metal binding could occur ( Supplementary Fig. S5). 580 Double mutant PglZ T538A/H741A was not stabilised by any metal indicating that metal binding in the 581 active site was no longer possible ( Supplementary Fig. S5 ). The double mutant was, however, still 582 destabilised by copper and zinc, suggesting effects for copper and zinc seen with both this mutant and 583 also PglZ WT are due to non-specific binding (Supplementary Fig. S5). The melting temperatures for 584 PglZ WT and mutants indicated that T538A reduces overall stability, but H741A has less of an impact 585 (Supplementary Fig. S5E). 586 587 Initial tests for potential phosphodiesterase activity using bis-pNPP as a substrate with PglZ 588 WT and additional zinc resulted in precipitation, and so magnesium, manganese, and calcium were all 589 tested as alternates, with manganese showing the greatest levels of activity (Supplementary Fig. S6). 590 Manganese was therefore selected as an alternate metal in bis-pNPP phosphodiesterase activity 591 assays. Having stripped metals from the samples and restored manganese, E. fergusonii PglZ WT 592 showed robust production of p-nitrophenol from bis-pNPP, at levels greater than for untreated PglZ 593 WT that contained the zinc remaining after purification (Fig. 3D). In contrast, when all mutants were 594 EDTA treated, re-purified, and then provided manganese, PglZ H741A had reduced activity, and both 595 PglZ T538A and the double mutant T538A/H741A lacked activity, demonstrating levels similar to those 596 observed for the PglZ WT sample that was without metal following EDTA treatment ( Fig. 3D). This 597

indicated that PglZ, like PorX, can cleave bis-pNPP in a metal -dependent manner, and that 598 mutations interfering with the likely metal binding site caused reduced activity. 599 600 Cleavage of cyclic oligonucleotides was then tested by incubating E. fergusonii PglZ WT with 601 cA6 and analysing the resulting products by high performance liquid chromatography (HPLC). Having 602 tested a range of metals it was noted that zinc was the preferred metal in these assays, and was used 603 at a reduced concentration to prevent protein destabilisation and precipitation. E. fergusonii PglZ WT 604 robustly linearised cA6 and sequentially cleaved nucleo tide products, indicated by a trace for each 605 linear species down to AMP ( Fig. 3E ). PglZ cleavage activity was ablated by EDTA treatment, and 606 mutant PglZ T538A showed no appreciable activity ( Fig. 3F). Mutant H741A appeared able to cleave 607 cA6 but did not produce further cleavage products ( Fig. 3F). PglZ was then tested against a broader 608 range of nucleotides ( Supplementary Figs . S7 and S8 ). PglZ was observed to cleave cyclic 609 oligonucleotides (cA4, cA3, cA2, and cG2) and linear oligonucleotides (pApA, pApG, pGpG, c(ApGp), 610 and c[A(35)pA(35)pG(35)]) containing both adenosine and guanosine, including an oligonucleotide 611 with 2-5 rather than 3 -5 phosphodiester linkages (c[G(25)pA(35)p]). PglZ was unable to cleave 612 cyclic mononucleotides (cAMP, cGMP, cTMP, cUMP, cCMP, 23 cAMP, 23 cGMP, and 23 cUMP) or 613 dinucleotide polyphosphates (Ap3A, Ap4A, and Ap4G) in the presence of Zn. Following this analysis 614 we returned to testing metal usage and noted that at low manganese concentrations we were also 615 able to observe PglZ -dependent cleavage of both cA6 and pApA ( Supplementary Fig . S7C). 616 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 22 Collectively, these data show robust metal -dependent cyclic and linear oligonucleotide cleavage by 617 PglZ from a BREX system. 618 619 PglZ is an endonuclease that can cleave dsDNA 620 Having established that PglZ has nuclease activity against oligonucleotides we were curious as to 621 whether PglZ could cleave dsDNA. Plasmid pSG483, a pUC19 derivative that can be easily prepared as 622 supercoiled (S), nicked (N), relaxed (R) or linear (L) dsDNA, was selected as a suitable substrate to be 623 tested against E. fergusonii PglZ. Initial assays indicated that manganese would be the preferred metal 624 in this context, but supplementing with zinc did allow some PglZ nuclease activity (Supplementary Fig. 625 S9A). Incubation of pSG483 with a titration of E. fergusonii PglZ WT revealed both nicking and 626 linearisation activities, which were metal-dependent and could be inhibited by ATP ( Fig. 4A). This 627 confirmed that PglZ can nick and cut dsDNA, and is an endonuclease. Due to the observed inhibition 628 by ATP, a range of mononucleotides were then tested. All NTPs, dNTPs and AMP-PNP inhibited PglZ 629 nuclease activity, but AMP did not (Supplementary Fig. S9B). The PglZ mutants were then tested for 630 nuclease activity (Fig. 4B). E. fergusonii PglZ H741A had increased nicking but decreased linearisation 631 activity, whereas PglZ T538A and the double mutant PglZ T538A/H741A were both ablated for activity 632 (Fig. 4B). This follows the previous observed trend for activity ( Fig. 3F) and indicates T538A prevents 633 metal binding and therefore activity, whilst H741A reduces metal binding and activity. 634 635 The BREX methyltransferase, PglX, determines sequence recognition for host methylation and 636 phage defence (17). E. fergusonii PglX recognises the sequence GCTAAT, and there is 1 copy of this 637 motif in pSG483. A mutant pSG483 was generated (pSG483BREX KO) with the GCTAAT motif mutated to 638 GCTATT to allow testing of whether PglZ cleavage is dependent on BREX motifs. When a titration of E. 639 fergusonii PglZ WT was titrated against pSG483 BREX KO there was no observed difference to the result 640 with pSG483 ( Figs. 4A and 4C ). We then considered whether BREX methylation might impact PglZ 641 nuclease activity. We prepared pBREXxl WT, a plasmid encoding the full E. fergusonii locus and the 642 mutant pBREXxl-ΔpglX. Each plasmid has previously been shown to be BREX methylated and lacking 643 methylation, respectively (24). E. fergusonii PglZ WT caused equal degradation of both plasmids, 644 indicating BREX methylation does not impact PglZ activity under these isolated conditions 645 (Supplementary Figs. S9C-D). 646 647 Phage Pau was shown to be susceptible to E. fergusonii BREX in an earlier study (24). Phage 648 Pau does not have modified DNA (23), unlike phage T4, which has modified cytosines and so is 649 inherently resistant to BREX. When tested, E. fergusonii PglZ was able to cleave genomic DNAs from 650 both these phages (Fig. 4D). The cleavage did not produce a distinct pattern, rather a faint smear of 651 products, demonstrating that PglZ is likely a sequence-independent endonuclease (Fig. 4D). It was also 652 interesting that T4 cytosine modifications did not impact PglZ cleavage when in isolation, though T4 653 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 23 654 655 656 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 24 (and other modified phages) are resistant to BREX phage defence. E. fergusonii PglZ WT could also 657 cause sequence -independent cleavage of ssDNA, using M18mp13 genomic DNA as substrate 658 (Supplementary Fig. S10A ). There was no activity, however, on MS2 phage genomic RNA 659 (Supplementary Fig. S10B). 660 661 Finally, to ensure our biochemical data and structural study are aligned, we confirmed that 662 Salmonella PglZ WT also demonstrated metal -dependent nicking and linearisation of pSG483 663 (Supplementary Fig. S10C ). Salmonella PglZ showed a preference for zinc or magnesium and in 664 contrast to E. fergusonii PglZ, Salmonella PglZ could also use calcium and copper , and could not use 665 manganese (Supplementary Fig. S10C). 666 667 PglZ:BrxB interactions can be ablated by interface mutations 668 As both E. fergusonii and Salmonella PglZ were shown to be nucleases, we also wanted to demonstrate 669 that E. fergusonii PglZ and BrxB also form sub -complexes as observed for the Salmonella and 670 Acinetobacter homologues (Figs. 1 and 2, Supplementary Fig. S1). Our hybrid model identified several 671 Salmonella BrxB residues important for the PglZ:BrxB interaction (Fig. 2B). A suite of E. fergusonii BrxB 672 WT and mutant proteins were expressed and purified ( Supplementary Fig. S11A ). None of the 673 proteins contained any metals after purification, as analysed by ICP -MS (Supplementary Fig. S11B). 674 SEC analysis of E. fergusonii BrxB WT showed that it formed both monomer and dimer peaks 675 (Supplementary Figs. S11C and D ), as confirmed by native mass spectrometric analysis 676 (Supplementary Fig. S11E). Analytical SEC was performed using E. fergusonii BrxB WT, PglZ and BrxB 677 WT with PglZ (Fig. 5A). Incubating BrxB WT with PglZ caused higher order complexes to form, as shown 678 by elution profiles and corresponding SDS-PAGE analysis of the peaks (Fig. 5A). This indicated that E. 679 fergusonii PglZ:BrxB sub-complexes were also forming, though perhaps with higher order forms being 680 produced beyond those observed for Salmonella homologues (Fig. 2A). 681 682 In contrast, co-incubation of E. fergusonii PglZ WT with BrxB R46A and BrxB W135A failed to produce 683 PglZ:BrxB complexes (Figs. 5B and C ). BrxB mutants E47A, E89A and S133A generated intermediate 684 elution profiles, indicating some complexes were forming, but to lesser extent than with BrxB WT 685 (Supplementary Fig. S12). Mass photometric analysis of E. fergusonii PglZ incubated with BrxB WT and 686 mutants showed similar trends, in that complexes formed with BrxB WT, none formed with BrxB R46A 687 or BrxB W135A, or BrxB S133A in these conditions, and complexes formed but less robustly with BrxB 688 E89A and BrxB E47A (Fig. 5D). 689 690 Finally, we also examined by analytical SEC whether any of the E. fergusonii PglZ mutants 691 T538A, H741A or T538A/H741A would impact BrxB WT binding and formation of higher order 692 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 25 693 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 26 complexes. As expected, due to these mutations being distant from the BrxB binding site (Figs. 2B and 694 3A), no impact on complex formation was observed (Supplementary Figure S13). Together, these data 695 support comparisons between our two model homologues, as both were shown to have nuclease 696 activity, and also now both have been shown to form sub-complexes. 697 698 Interaction with BrxB impacts neither PglZ nuclease activity nor inhibition of 699 nuclease activity by ATP 700 Gel-based nuclease activities were then used to investigate the impact of BrxB interactions on PglZ 701 activity. In these assays, E. fergusonii BrxB had no identifiable nicking or linearisation activity (Fig. 6A). 702 Titration of E. fergusonii BrxB against E. fergusonii PglZ had no appreciable impact on PglZ nicking and 703 linearisation activities until the highest BrxB concentration ( Fig. 6A). Comparisons of the AlphaFold 704 model for the BrxB structure using DALI (38) indicated some similarity to nucleotide binding regions 705 of AAA+ proteins, but BrxB appears to be lacking key Walker motif residues. As BrxB had the potential 706 for binding ATP, we investigated whether BrxB might impact the observed inhibition of PglZ activity 707 by ATP (Fig. 4A). When the same PglZ to BrxB titration was performed in the presence of ATP there 708 was no indication that BrxB could overcome inhibition of PglZ activity by ATP (Fig. 6A ). For 709 completeness, we also tested whether non -interacting E. fergusonii BrxB mutants impacted PglZ 710 activity, but no effect was observed (Fig. 6B). Next, we used HPLC analysis of oligonucleotide cleavage 711 as another measure of BrxB impact. Neither BrxB WT nor BrxB R46A altered the ability of E. fergusonii 712 PglZ to cleave cA6 or pApA ( Fig. 6C). This indicates that the role of BrxB, at least in these isolated 713 conditions, is independent of PglZ nuclease activity, and likely has more relevance in the context of 714 larger BREX complexes. 715 716 717 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 27 718 719 720 721 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 28 PglZ nuclease activity contributes to BREX phage defence but not BREX -722 dependent methylation 723 With the E. fergusonii PglZ and BrxB mutations now characterised biochemically we examined their 724 impact on the two measurable BREX phenotypes, phage defence and BREX -dependent methylation. 725 Mutations were constructed in the context of pBrxXL, which encodes the full E. fergusonii BREX locus 726 under native promoters (24). The suite of mutants were tested for defence against phage Pau from 727 the Durham collection (23), measured by Efficiency of Plating (EOP), using an appropriate vector 728 control. The positive and negative controls pBrxXL and pBrxXL -ΔpglX provided strong and no phage 729 defence, respectively ( Fig. 7A). BrxB mutant constructs S133A and W135A, and the PglZ H741A 730 construct all showed a small reduction in phage defence activity, around 10-fold (Fig. 7A). PglZ T538A 731 and double mutant T538A/H741A constructs showed a large reduction in defence of around 3 logs, 732 but remained impressively active ( Fig. 7A). These data indicate that mutations preventing PglZ:BrxB 733 interactions or ablating PglZ nuclease activity have an impact but can be compensated for in vivo. 734 735 Genomic DNA was extracted from each strain and PacBio sequencing analysis was performed 736 to investigate BREX-dependent methylation, examining N6mA methylation on the fifth adenine of the 737 GCTAAT motif ( Fig. 7B). Positive control pBrxXL WT had 99.4 % methylation of BREX motifs, and 738 negative control pBrxXL-ΔpglX had no detectable methylation, as observed previously (24). In all cases, 739 the current mutations in pBrxXL had no impact on BREX-dependent methylation (Fig. 7B), indicating 740 PglZ nuclease activity works independently from BREX methylation. 741 742 743 744 745 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 29 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 30

761 Type I BREX systems perform phage defence and BREX-dependent methylation, but the mechanisms 762 for each activity have proven difficult to uncover. Having previously performed individual studies of 763 BREX components and regulators (17–22), in this study we examined what higher order BREX 764 complexes form in cells and focussed on characterisation of PglZ, BrxB and the resulting stable 765 PglZ:BrxB sub-complex. 766 767 Using Salmonella BrxB as bait we observed pull -down of BrxB with BrxC, PglX, PglZ and also 768 BrxL ( Fig. 1 ). Although BrxL is the effector protein needed for phage defence with the E. coli and 769 Acinetobacter BREX systems (20, 39), we previously noted that Salmonella BREX can provide phage 770 defence without BrxL (23), and there is potential for BrxL having a regulatory role in this system (40). 771 It was therefore curious that BrxL readily associates with other Salmonella BREX components . 772 However, low levels of BrxL can also be seen in similar pull-down experiments performed using E. coli 773 BREX (18). Using BrxL as bait does also pull -down other BREX components, though the data do show 774 that the more robust complex appears to be formed of BrxB, BrxC, PglX and PglZ (18). We noted strong 775 association between PglZ and BrxB, also seen with E. coli BREX (18), and so chose to examine this sub-776 complex as a step towards understanding the structure and function of higher order BREX complexes. 777 778 Our resulting hybrid model of Salmonella PglZ:BrxB (Fig. 2) uses 4.45 Å resolution density 779 obtained by cryoEM to fit AlphaFold (33) predicted domains of each protein. Notably, our model 780 demonstrates points of structural flexibility that allow movement of the PglZ:BrxB interaction domain 781 and the PglZ C-terminal β-barrel domain (Supplementary Fig. S3). This flexibility is the likely cause of 782 the limited resolution observed for the corresponding cryoEM analysis. 783 784 The first evidence of biochemical activity for PglZ domains, originally considered an alkaline 785 phosphatase (15), came from demonstration that the PglZ domain of PorX , a two -component 786 signalling system response regulator, could act as a phosphodiesterase and linearise cyclic nucleotides 787 (16). PorX activity i s zinc -dependent. Having examined the structure of Salmonella PglZ:BrxB, we 788 switched to E. fergusonii PglZ and BrxB for biochemical characterisation as the proteins behaved more 789 reproducibly under assay conditions. After substantial efforts to identify a preferred metal and 790 concentration, it could be demonstrated that E. fergusonii PglZ has similar activity to PorX, cleaving a 791 wide range of cyclic nucleotides, but not cyclic mononucleotides or dinucleotide polyphosphates (Fig. 792 3 and Supplementary Fig. S4-8). 793 794 We hypothesised that our observed E. fergusonii PglZ activity could impact dsDNA. When 795 tested, E. fergusonii PglZ nicked and linearised dsDNA ( Fig. 4). This activity was applicable to ssDNA, 796 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 31 but not dsRNA, and was independent of BREX motifs, appeared sequence -independent, was not 797 impacted by BREX methylation and was not impacted by larger DNA modifications such as glucose 798 modifications to hydroxymethylated cytosines in phage T4 genomic DNA (Fig. 4 and Supplementary 799 Figs. S9 and S10). Initial investigations of BREX activity indicated little digestion of invading phage DNA, 800 merely an inhibition of phage genome replication, making BREX a classic “restriction” system (15). 801 Nicking is a hallmark of multiple other phage defence systems, including Shedu, Lamassu, Dnd and 802 Gabija (41–45). In the latter case, Gabija activity is also regulated by the detection and degration of 803 nucleotides (45). As we observed inhibition of PglZ nicking activity in the presence of ATP (Fig. 4), we 804 cannot rule out analogous regulatory activity by PglZ in the context of a full BREX mechanism . It is 805 unclear whether ATP might be competing for the PglZ domain catalytic site, or have another inhibitory 806 binding site that alters activity. We also cannot dismiss a potential role for PglZ-dependent nicking in 807 protecting from invading DNA, perhaps as a precursor licensing step to allow further inhibition of 808 replication. 809 810 The role of BrxB has thus far remained hypothetical, yet the predicted fold mimics AAA+ 811 nucleotide binding domains. We were able to demonstrate binding of E. fergusonii BrxB to PglZ, and 812 pinpoint residues required for stable complex formation (Fig. 5 and Supplementary Fig. S12 and S13). 813 When we then tested whether BrxB might bind and therefore alter the observed inhibition of PglZ by 814 ATP, nothing changed (Fig. 6). Alternatively, and due to the observed data in Salmonella (Fig. 1) and 815 E. coli (18), we postulate that the role of BrxB might be as a scaffold protein, participating within and 816 allowing connections between multiple BREX components within higher order complexes. 817 818 When assaying the two phenotypes of phage defence and methylation we saw that ablation 819 of PglZ:BrxB interactions made only a small impact on phage defence (Fig. 7A), perhaps because within 820 a higher order complex other interactions occur to support function. In contrast, removal of PglZ 821 nuclease activity by mutation had a stronger impact though still did not remove all phage defence 822 activity, indicating that the overall BREX mechanism can compensate for loss or reduction in PglZ 823 function (Fig. 7A). Whilst deletion of pglZ prevents BREX phage defence and methylation (17, 20, 39), 824 our mutations of either PglZ nuclease activity or BrxB binding had no impact on methylation (Fig. 7B), 825 indicating PglZ likely plays a role in formation of the BREX methylation complex, but not in that specific 826 activity. 827 828 A working model for BREX activity was recently posited, wherein a BREX-BCXZ complex would 829 form and move along DNA to allow methylation (18). Our data support formation of this complex (Fig. 830 1), and contrary to data indicating PglX alone is sufficient for methylation (46), we could not see 831 methylation using Salmonella or the same E. coli homologue (17, 18)and expression of PglX alone in 832 cells also does not result in methylation (39). A BREX -BCXZ complex would have to be able to 833 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 32 distinguish between DNA templates containing BREX methylation on one strand following host 834 genome replication, and target invading DNA containing no BREX methylation. When invading DNA is 835 recognised, we envision a role for the PglZ nuclease in which nicking of target DNA licenses the BREX-836 BCXZ complex to switch from methylation surveillance to restriction. This could include recruitment 837 of BrxL and, for instance, movement of the BREX complex to stall replication forks. Nevertheless, there 838 remains many questions as to the specifics of BREX activity and our data indicate clear next steps in 839 the characterisation of higher order BREX complexes. 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 33

861 This research was supported by the Electron Microscopy Shared Resource, RRID:SCR_022611, of the 862 Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium (P30 CA015704). CryoEM 863 molecular graphics and analyses were performed with UCSF ChimeraX, developed by the Resource for 864 Biocomputing, Visualization, and Informatics at the University of California, San Francisco, with 865 support from National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and 866 Computational Biology, National Institute of Allergy and Infectious Diseases. 867 868 Author contributions: J.J.R. and L.A.D. contributed equally. J.J.R. expressed all E. fergusonii 869 proteins and performed biochemical analyses. L.A.D. performed cryoEM practical aspects, data 870 collection and data processing. M.P. produced Salmonella proteins for biochemistry and performed 871 co-expression analyses and phage assays. A.K. performed in vivo pull-down analyses and mass 872 photometry. A.N. performed PacBio sequencing and analysis. A.J.K., S.M. and J.P-A. produced and 873 analysed Acinetobacter proteins. T.R.B. produced Salmonella protein for cryoEM. D.L.S., B.L.S., B.K.K., 874 and T.R.B. supervised the project and obtained funding. All authors contributed to data analysis and 875 writing the manuscript. 876 877 SUPPLEMENTARY DATA 878 Supplementary data have been provided and comprise 13 Figures and one Table. 879 880 CONFLICT OF INTEREST 881 T.R.B. is an employee of, and B.L.S. is a paid consultant for, New England Biolabs, which provided 882 funding support for this study and which develops a wide variety of phage restriction systems for 883 commercial sale. 884 885 FUNDING 886 This work was supported by a Biotechnology and Biological Sciences Research Council Newcastle -887 Liverpool-Durham Doctoral Training Partnership studentship [grant number BB/T008695/1] to J.J.R., 888 a Biotechnology and Biological Sciences Research Council responsive mode grant [grant number 889 BB/Y003659/1] to M.P., a Lister Institute Prize Fellowship to A.K. and T.R.B., New England Biolabs 890 (NEB), the Fred Hutchinson Cancer Center (FHCC) and the NIH for both BLS (R01 GM105691) and BKK 891 (R15 GM140375). 892 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 34 893 For the purpose of open access, the authors have applied a CC BY public copyright licence to 894 any Author Accepted Manuscript version arising from this submission. 895 896 DATA AVAILABILITY 897 The cryoEM model and corresponding maps for PglZ complexed with BrxB have been deposited in the 898 RCSB PDB database (ID code 9NV3) and in the EMDB (ID code EMD -49827). All other data needed to 899 evaluate the conclusions in the paper are present in the paper and/or Supplementary Data. 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 35

919 1. Tesson,F., Hervé,A., Mordret,E., Touchon,M., d’Humières,C., Cury,J. and Bernheim,A. (2022) 920 Systematic and quantitative view of the antiviral arsenal of prokaryotes. Nat Commun, 13, 921 2561. 922 2. Labrie,S.J., Samson,J.E. and Moineau,S. (2010) Bacteriophage resistance mechanisms. Nat Rev 923 Microbiol, 8, 317–327. 924 3. Hampton,H.G., Watson,B.N.J. and Fineran,P.C. (2020) The arms race between bacteria and their 925 phage foes. Nature, 577, 327–336. 926 4. Murtazalieva,K., Mu,A., Petrovskaya,A. and Finn,R.D. (2024) The growing repertoire of phage anti-927 defence systems. Trends Microbiol, 32. 928 5. Georjon,H. and Bernheim,A. (2023) The highly diverse antiphage defence systems of bacteria. 929 Nature Reviews Microbiology 2023 21:10, 21, 686–700. 930 6. Doron,S., Melamed,S., Ofir,G., Leavitt,A., Lopatina,A., Keren,M., Amitai,G. and Sorek,R. (2018) 931 Systematic discovery of antiphage defense systems in the microbial pangenome. Science 932 (1979), 359, eaar4120. 933 7. Vassallo,C.N., Doering,C.R., Littlehale,M.L., Teodoro,G.I.C. and Laub,M.T. (2022) A functional 934 selection reveals previously undetected anti-phage defence systems in the E. coli pangenome. 935 Nat Microbiol, 7, 1568–1579. 936 8. Macdonald,E., Wright,R., Connolly,J.P.R., Strahl,H., Brockhurst,M., van Houte,S., Blower,T.R., 937 Palmer,T. and Mariano,G. (2023) The novel anti-phage system Shield co-opts an RmuC domain 938 to mediate phage defense across Pseudomonas species. PLoS Genet, 19, e1010784. 939 9. Cummins,T., Songra,S., Garrett,S., Blower,T.R. and Mariano,G. (2024) Multi-conflict islands are a 940 widespread trend within Serratia spp. Cell Rep. 941 10. Mariano,G. and Blower,T.R. (2023) Conserved domains can be found across distinct phage 942 defence systems. Mol Microbiol, 10.1111/mmi.15047. 943 11. Wein,T. and Sorek,R. (2022) Bacterial origins of human cell-autonomous innate immune 944 mechanisms. Nat Rev Immunol, 10.1038/s41577-022-00705-4. 945 12. Chinenova,T.A., Mkrtumian,N.M. and Lomovskaia,N.D. (1982) Genetic characteristics of a new 946 phage resistance trait in Streptomyces coelicolor A3(2). Genetika, 18, 1945–1952. 947 13. Sumby,P. and Smith,M.C.M. (2002) Genetics of the phage growth limitation (Pgl) system of 948 Streptomyces coelicolor A3(2). Mol Microbiol, 44, 489–500. 949 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 36 14. Hoskisson,P.A., Sumby,P. and Smith,M.C.M. (2015) The phage growth limitation system in 950 Streptomyces coelicolor A(3)2 is a toxin/antitoxin system, comprising enzymes with DNA 951 methyltransferase, protein kinase and ATPase activity. Virology, 477, 100–109. 952 15. Goldfarb,T., Sberro,H., Weinstock,E., Cohen,O., Doron,S., Charpak-Amikam,Y., Afik,S., Ofir,G. and 953 Sorek,R. (2015) BREX is a novel phage resistance system widespread in microbial genomes. 954 EMBO J, 34, 169–83. 955 16. Schmitz,C., Madej,M., Nowakowska,Z., Cuppari,A., Jacula,A., Ksiazek,M., Mikruta,K., 956 Wisniewski,J., Pudelko-Malik,N., Saran,A., et al. (2022) Response regulator PorX coordinates 957 oligonucleotide signalling and gene expression to control the secretion of virulence factors. 958 Nucleic Acids Res, 10.1093/nar/gkac1103. 959 17. Went,S.C., Picton,D.M., Morgan,R.D., Nelson,A., Brady,A., Mariano,G., Dryden,D.T.F., Smith,D.L., 960 Wenner,N., Hinton,J.C.D., et al. (2024) Structure and rational engineering of the PglX 961 methyltransferase and specificity factor for BREX phage defence. Nat Commun, 15. 962 18. Drobiazko,A., Adams,M.C., Skutel,M., Potekhina,K., Kotovskaya,O., Trofimova,A., Matlashov,M., 963 Yatselenko,D., Maxwell,K.L., Blower,T.R., et al. (2025) Molecular basis of foreign DNA 964 recognition by BREX anti-phage immunity system. Nat Commun, 16, 1825. 965 19. Picton,D.M., Harling-Lee,J.D., Duffner,S.J., Went,S.C., Morgan,R.D., Hinton,J.C.D. and Blower,T.R. 966 (2022) A widespread family of WYL-domain transcriptional regulators co-localizes with diverse 967 phage defence systems and islands. Nucleic Acids Res, 50, 5191–5207. 968 20. Luyten,Y.A., Hausman,D.E., Young,J.C., Doyle,L.A., Higashi,K.M., Ubilla-Rodriguez,N.C., 969 Lambert,A.R., Arroyo,C.S., Forsberg,K.J., Morgan,R.D., et al. (2022) Identification and 970 characterization of the WYL BrxR protein and its gene as separable regulatory elements of a 971 BREX phage restriction system. Nucleic Acids Res, 50, 5171–5190. 972 21. Beck,I.N., Picton,D.M. and Blower,T.R. (2022) Crystal structure of the BREX phage defence 973 protein BrxA. Curr Res Struct Biol, 4, 211–219. 974 22. Shen,B.W., Doyle,L.A., Werther,R., Westburg,A.A., Bies,D.P., Walter,S.I., Luyten,Y.A., 975 Morgan,R.D., Stoddard,B.L. and Kaiser,B.K. (2023) Structure, substrate binding and activity of a 976 unique AAA+ protein: the BrxL phage restriction factor. Nucleic Acids Res, 977 10.1093/nar/gkad083. 978 23. Kelly,A., Went,S.C., Mariano,G., Shaw,L.P., Picton,D.M., Duffner,S.J., Coates,I., Herdman-Grant,R., 979 Gordeeva,J., Drobiazko,A., et al. (2023) Diverse Durham collection phages demonstrate 980 complex BREX defense responses. Appl Environ Microbiol, 89. 981 24. Picton,D.M., Luyten,Y.A., Morgan,R.D., Nelson,A., Smith,D.L., Dryden,D.T.F., Hinton,J.C.D. and 982 Blower,T.R. (2021) The phage defence island of a multidrug resistant plasmid uses both BREX 983 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 37 and type IV restriction for complementary protection from viruses. Nucleic Acids Res, 49, 984 11257–11273. 985 25. Anton,B.P., Mongodin,E.F., Agrawal,S., Fomenkov,A., Byrd,D.R., Roberts,R.J. and Raleigh,E.A. 986 (2015) Complete Genome Sequence of ER2796, a DNA Methyltransferase-Deficient Strain of 987 Escherichia coli K-12. PLoS One, 10, e0127446. 988 26. Beck,I.N., Arrowsmith,T.J., Grobbelaar,M.J., Bromley,E.H.C., Marles-Wright,J. and Blower,T.R. 989 (2024) Toxin release by conditional remodelling of ParDE1 from Mycobacterium tuberculosis 990 leads to gyrase inhibition. Nucleic Acids Res, 52, 1909–1929. 991 27. Cai,Y., Usher,B., Gutierrez,C., Tolcan,A., Mansour,M., Fineran,P.C., Condon,C., Neyrolles,O., 992 Genevaux,P. and Blower,T.R. (2020) A nucleotidyltransferase toxin inhibits growth of 993 Mycobacterium tuberculosis through inactivation of tRNA acceptor stems. Sci Adv, 6, 994 eabb6651. 995 28. Grøftehauge,M.K., Hajizadeh,N.R., Swann,M.J. and Pohl,E. (2015) Protein-ligand interactions 996 investigated by thermal shift assays (TSA) and dual polarization interferometry (DPI). Acta 997 Crystallogr D Biol Crystallogr, 71, 36–44. 998 29. Tegunov,D. and Cramer,P. (2019) Real-time cryo-electron microscopy data preprocessing with 999 Warp. Nature Methods 2019 16:11, 16, 1146–1152. 1000 30. Punjani,A., Rubinstein,J.L., Fleet,D.J. and Brubaker,M.A. (2017) cryoSPARC: algorithms for rapid 1001 unsupervised cryo-EM structure determination. Nature Methods 2017 14:3, 14, 290–296. 1002 31. Punjani,A., Zhang,H. and Fleet,D.J. (2020) Non-uniform refinement: adaptive regularization 1003 improves single-particle cryo-EM reconstruction. Nature Methods 2020 17:12, 17, 1214–1221. 1004 32. Goddard,T.D., Huang,C.C., Meng,E.C., Pettersen,E.F., Couch,G.S., Morris,J.H. and Ferrin,T.E. 1005 (2018) UCSF ChimeraX: Meeting modern challenges in visualization and analysis. Protein 1006 Science, 27, 14–25. 1007 33. Abramson,J., Adler,J., Dunger,J., Evans,R., Green,T., Pritzel,A., Ronneberger,O., Willmore,L., 1008 Ballard,A.J., Bambrick,J., et al. (2024) Accurate structure prediction of biomolecular interactions 1009 with AlphaFold 3. Nature 2024 630:8016, 630, 493–500. 1010 34. Liebschner,D., Afonine,P. V., Baker,M.L., Bunkoczi,G., Chen,V.B., Croll,T.I., Hintze,B., Hung,L.W., 1011 Jain,S., McCoy,A.J., et al. (2019) Macromolecular structure determination using X-rays, 1012 neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol, 75, 861–1013 877. 1014 35. Kim,D.N., Moriarty,N.W., Kirmizialtin,S., Afonine,P. V., Poon,B., Sobolev,O. V., Adams,P.D. and 1015 Sanbonmatsu,K. (2019) Cryo_fit: Democratization of flexible fitting for cryo-EM. J Struct Biol, 1016 208, 1–6. 1017 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 38 36. Afonine,P. V., Poon,B.K., Read,R.J., Sobolev,O. V., Terwilliger,T.C., Urzhumtsev,A. and Adams,P.D. 1018 (2018) Real-space refinement in PHENIX for cryo-EM and crystallography. Acta Crystallogr D 1019 Struct Biol, 74, 531–544. 1020 37. Krissinel,E. and Henrick,K. (2007) Inference of Macromolecular Assemblies from Crystalline State. 1021 J Mol Biol, 372, 774–797. 1022 38. Holm,L. (2022) Dali server: structural unification of protein families. Nucleic Acids Res, 50, W210–1023 W215. 1024 39. Gordeeva,J., Morozova,N., Sierro,N., Isaev,A., Sinkunas,T., Tsvetkova,K., Matlashov,M., 1025 Truncaite,L., Morgan,R.D., Ivanov,N. V, et al. (2019) BREX system of Escherichia coli 1026 distinguishes self from non-self by methylation of a specific DNA site. Nucleic Acids Res, 47, 1027 253–265. 1028 40. Zaworski,J., Dagva,O., Brandt,J., Baum,C., Ettwiller,L., Fomenkov,A. and Raleigh,E.A. (2022) 1029 Reassembling a cannon in the DNA defense arsenal: Genetics of StySA, a BREX phage exclusion 1030 system in Salmonella lab strains. PLoS Genet, 18, e1009943. 1031 41. Loeff,L., Walter,A., Rosalen,G.T. and Jinek,M. (2025) DNA end sensing and cleavage by the Shedu 1032 anti-phage defense system. Cell, 188, 721-733.e17. 1033 42. Loeff,L., Adams,D.W., Chanez,C., Stutzmann,S., Righi,L., Blokesch,M. and Jinek,M. (2024) 1034 Molecular mechanism of plasmid elimination by the DdmDE defense system. Science, 385, 1035 188–194. 1036 43. Tang,Y., Wu,D., Zhang,Y., Liu,X., Chu,H., Tan,Q., Jiang,L., Chen,S., Wu,G. and Wang,L. (2024) 1037 Molecular basis of the phosphorothioation-sensing antiphage defense system IscS-DndBCDE-1038 DndI. Nucleic Acids Res, 52, 13594–13604. 1039 44. Cheng,R., Huang,F., Wu,H., Lu,X., Yan,Y., Yu,B., Wang,X. and Zhu,B. (2021) A nucleotide-sensing 1040 endonuclease from the Gabija bacterial defense system. Nucleic Acids Res, 49, 5216–5229. 1041 45. Cheng,R., Huang,F., Lu,X., Yan,Y., Yu,B., Wang,X. and Zhu,B. (2023) Prokaryotic Gabija complex 1042 senses and executes nucleotide depletion and DNA cleavage for antiviral defense. Cell Host 1043 Microbe, 31, 1331-1344.e5. 1044 46. Li,S., Xu,T., Meng,X., Yan,Y., Zhou,Y., Duan,L., Tang,Y., Zhu,L. and Sun,L. (2024) Ocr-mediated 1045 suppression of BrxX unveils a phage counter-defense mechanism. Nucleic Acids Res, 52, 8580–1046 8594. 1047 1048 1049 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 39 FIGURE LEGENDS 1050 Figure 1. BREX proteins form higher order complexes. ( A) Schematic of the BREX phage defence 1051 islands from Salmonella and Escherichia fergusonii. The Salmonella island encodes BREX and PARIS 1052 (ariA, ariB) defence systems. The E. fergusonii island encodes BREX and a Type IV restriction enzyme 1053 of the GmrSD family, BrxU. (B) Pull-down of Salmonella BREX complexes. E. coli BL21 (DE3) pRARE 1054 was transformed with an inducible plasmid expressing His-strep tagged BrxB (or vector control), and a 1055 second plasmid expressing the six BREX genes brxA, brxB, brxC, pglX, pglZ and brxL (or vector 1056 control). Pull-down samples were analysed by SDS-PAGE and indicated bands were identified by mass 1057 spectrometry. (C) Expression of Salmonella BREX proteins in pairs from pET DUET -based vectors, 1058 pulled down with His -SUMO-BrxB. Pull-down samples were analysed on SDS -PAGE and indicated 1059 bands were identified by mass spectrometry. ( D) Toxicity during expression of Salmonella BREX 1060 proteins, measured as viable counts. Error bars represent the standard deviation of the mean from 1061 triplicate data. 1062 1063 Figure 2. Structure of the Salmonella PglZ:BrxB stable sub -complex. ( A) Size exclusion 1064 chromatography traces of independent Salmonella PglZ and BrxB purifications, and the PglZ:BrxB co -1065 expression sample used for cryoEM, show that PglZ and BrxB form a stable complex. ( B) A cryoEM 1066 map generated from single particles of the Salmonella PglZ:BrxB complex validates the interaction 1067 between both proteins, as well as the location of the PglZ:BrxB interaction surfaces and identity of 1068 residues in the protein -protein interface (inset box). The resulting model is closely related to the 1069 corresponding AlphaFold prediction for the Salmonella PglZ:BrxB complex, albeit with slight 1070 rearrangements corresponding to a small rotation of the N-terminal domain of PglZ and associated BrxB 1071 relative to the larger core of PglZ (Supplementary Figure S3). 1072 1073 Figure 3. PglZ can cleave cyclic nucleotides in a metal-dependent manner. (A) Overlay of AlphaFold2 1074 E. fergusonii PglZ predicted structure with PorX from Porphyromonas gingivalis (PDB: 7PVK). (B) ICP-1075 MS of PglZ WT and mutants showing divalent cations bound following purification. Plotted data 1076 represent mean values ± SD. Metal content is plotted as a percentage of the total protein in the sample. 1077 (C) Thermal shift assays performed upon PglZ WT (5 μM) following incubation with EDTA or metals 1078 (0.5 mM). Mean changes in melting temperature ( ΔTm) are plotted by comparison to PglZ WT in the 1079 absence of EDTA or metal, which is set as ‘0’. Error bars represent standard deviation (6 replicates) 1080 (D) Bis(4-nitrophenyl) phosphate (2.5 mM) phosphodiesterase assays using PglZ WT (2 μM) in the 1081 presence and absence of EDTA and Mn, and mutants in the presence of Mn. The Mn is supplied by 0.5 1082 mM MnCl2. Plotted data represent mean values ± SD (9 replicates). Absorbance (A405nm) represents the 1083 amount of reaction product p-nitrophenyl phosphate. ( E) HPLC analysis of cyclic hexa -adenosine 1084 monophosphate (cA6) cleavage by PglZ WT (2 μM) in the presence of Zn (10 μM). (F) HPLC analysis 1085 of cyclic hexa-adenosine monophosphate (cA6) cleavage by PglZ WT (2 μM) treated with EDTA, and 1086 mutants T538A and H741A in the presence of Zn (10 μM). Control reactions are represented by cA6 1087 alone, and standard mixes are comprised of pApA and AMP. All nucleotides in the reaction mixes are 1088 at 10 μM. Presented traces are representative of triplicate data. 1089 1090 Figure 4. PglZ is a metal-dependent nuclease that does not recognise BREX sites. (A) PglZ nicks and 1091 linearises supercoiled plasmid pSG483 in a metal -dependent manner that is inhibited by ATP. ( B) 1092 Mutation of metal binding site alters nuclease activity, either by shifting activity away from linearisation 1093 and towards nicking (H741A) or eliminating activity (T538A and T538A/H741A). ( C) PglZ can linearise 1094 plasmid pSG483 DNA with a mutated BREX site. (D) PglZ does not appear to have site-specific nicking 1095 activity on linearised phage DNA, nor be impacted by DNA modifications such as those found in T4. 1096 PglZ was titrated against constant supercoiled pSG483 plasmid DNA (6 nM) or phage gDNA (200 ng) 1097 in the presence and absence of MnCl2 (0.5 mM) and ATP (1 mM). Control lanes represent supercoiled 1098 (S), nicked (N), linear (L), and relaxed (multiple topoisomers; R) plasmid DNA, or contain the 1099 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint 40 appropriate DNA in the absence of protein. Assays are presented on 1.4% (w/v) or 0.8% (w/v) agarose 1100 1x TAE gels post -stained with ethidium bromide. Assays shown are representative of triplicate 1101 experiments. Data points and error bars represent the mean ± SD of triplicate data. 1102 1103 Figure 5. E. fergusonii BrxB mutants fail to form complexes with PglZ. Analytical SEC (S200i) and SDS-1104 PAGE analysis of PglZ:BrxB complexes formed with BrxB WT ( A), BrxB R46A (B), and BrxB W135A (C). 1105 Samples of PglZ (10 μM) with equimolar BrxB WT and mutants were made up to 100 μl and pre-1106 incubated for 15 min prior to loading on the S200i. The expected elution volumes (Ve) of various complex 1107 conformations are highlighted by black or red dotted lines, and the elution profile of PglZ incubated with 1108 BrxB is shown as a dark green solid line. Control elution profiles of PglZ alone (light green dashed line) 1109 and BrxB alone (dark blue dashed line) are also shown. Fractionated peak samples were resolved on 1110 15% (v/v) polyacrylamide gels for 1 h 15 min in tris -glycine running buffer and stained with Quick 1111 Coomassie. Protein identities are highlighted with black arrows. (D) Mass photometry of PglZ with BrxB 1112 WT and mutants. Counts were acquired for 60 s with BrxB (5 nM) or samples of PglZ (5 nM) pre -1113 incubated with equimolar BrxB WT and mutants in phosphate buffered saline (PBS). 1114 1115 Figure 6. BrxB interacting with PglZ does not appreciably alter PglZ nuclease activity. ( A) Incubation 1116 of PglZ (768 nM) with a titration of BrxB WT in the absence or presence of MnCl2 (0.5 mM) and ATP (1 1117 mM). (B) Incubation of PglZ WT (768 nM) with a titration of BrxB mutants R46A and W135A in the 1118 absence or presence of MnCl2 (0.5 mM). Control lanes represent supercoiled (S), nicked (N), linear (L), 1119 and relaxed (multiple topoisomers; R) plasmid DNA, or contain DNA in the absence of protein. Assays 1120 are presented on 1.4% (w/v) agarose 1x TAE gels post -stained with ethidium bromide. Assays shown 1121 are representative of triplicate experiments. (C) PglZ (2 μM) incubated in the presence of BrxB (10 μM) 1122 does not prevent cleavage of cA6 or pApA (10 μM). Control reactions are comprised of the nucleotide 1123 in the absence of protein. Standard mixes are comprised of pApA and AMP. Presented traces are 1124 representative of triplicate data. 1125 1126 Figure 7. PglZ:BrxB mutants have small impact on BREX phage defence and no impact on BREX 1127 methylation. (A) EOP results of phage Pau tested against E. fergusonii BREX constructs. Error bars 1128 represent standard deviation from the mean of triplicate data. ( B) PacBio sequencing results showing 1129 the percentage of BREX motif methylation in each strain. 1130 1131 .CC-BY 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted March 26, 2025. ; https://doi.org/10.1101/2025.03.26.645558doi: bioRxiv preprint