Single phage proteins sequester signals from TIR and cGAS-like enzymes

Abstract

Prokaryotic anti-phage immune systems use TIR and cGAS-like enzymes to produce 1''-3'-glycocyclic ADP-ribose (1''-3'-gcADPR) and cyclic dinucleotide (CDN) and cyclic trinucleotide (CTN) signalling molecules, respectively, which limit phage replication^1-3. However, how phages neutralize these distinct and common systems is largely unclear. Here we show that the Thoeris anti-defence proteins Tad1⁴ and Tad2⁵ both achieve anti-cyclic-oligonucleotide-based anti-phage signalling system (anti-CBASS) activity by simultaneously sequestering CBASS cyclic oligonucleotides. Apart from binding to the Thoeris signals 1''-3'-gcADPR and 1''-2'-gcADPR, Tad1 also binds to numerous CBASS CDNs and CTNs with high affinity, inhibiting CBASS systems that use these molecules in vivo and in vitro. The hexameric Tad1 has six binding sites for CDNs or gcADPR, which are independent of the two high-affinity binding sites for CTNs. Tad2 forms a tetramer that also sequesters various CDNs in addition to gcADPR molecules, using distinct binding sites to simultaneously bind to these signals. Thus, Tad1 and Tad2 are both two-pronged inhibitors that, alongside anti-CBASS protein 2 (Acb2^6-8), establish a paradigm of phage proteins that use distinct binding sites to flexibly sequester a considerable breadth of cyclic nucleotides.

Extended Data Figure 1.. Bacterial hosts that Tad-encoding phage likely infect contain multiple CBASS CD-NTases that produce cyclic oligonucleotides.

Bacteria from the genus (a) Clostridium, (b) Bacteroides, (c) Sphingobacterium, and (d) Bacillus cereus group contain CBASS CD-NTases that produce cyclic oligonucleotides (c-oligos) with strong, weak, or no binding affinity to the Tad proteins tested in this study. The relative frequency of CD-NTases is quantified as the number of CD-NTase from a specific clade divided by the total number of CD-NTases identified using the NCBI blastp (see Methods for details). The legend indicates the c-oligos that are known or predicted to be produced by the indicated CBASS CD-NTases (Whiteley et al., 2019; Ye et al., 2020; Morehouse et al., 2020; Fatma et al., 2021). CD-NTases with currently unknown nucleotide products are indicated in the legend and the graphs.

Extended Data Figure 2.. Structural comparison between CbTad1 and CmTad1.

a, The hexameric form of CbTad1 are shown in cartoon model. Left: CbTad1 (PDB code: 7UAV). Right: CbTad1–1”-2’ gcADPR (PDB code: 7UAW). b, Structural comparison between CbTad1 and CmTad1 hexamers. Left: structural comparison between apo CbTad1 (colored as in a) and apo CmTad1 (colored orange). Right: structural comparison between CbTad1–1”-2’ gcADPR (colored as in a) and apo CmTad1 (colored orange). c, Detailed binding in the hexamer interface of CbTad1. Residues involved in hexamer formation are shown as sticks. Red dashed lines represent polar interactions. d, Sequence alignment between Tad1 homologs. The cyclic trinucleotide (CTN) and cyclic dinucleotides (CDN)/gcADPR binding sites are marked in yellow and red, respectively. Representative sequences were intentionally selected to show CDN/CTN binding site mutations. Residues involved in hexamer formation are marked with blue triangles. e, SLS studies of purified CbTad1 and its E100A/E101A/M104A/W105A/R108A mutant. Calculated molecular weight is shown above the peaks. Profile of the molecular standard sample is colored in dark blue. f, Structural comparison between the dimer form of CbTad1 and CmTad1. Apo CbTad1 (PDB code: 7UAV) is colored green. Apo CmTad1 is colored in pink and magenta for the two protomers.

Extended Data Figure 3.. Tad1 binds to 2’,3’-/3’,2’-cGAMP using the same binding pocket as gcADPR molecules.

a, Overall structure of CbTad1 hexamer bound to 2’,3’-cGAMP, which is shown as yellow sticks. b, Structural superimposition of apo, 1”-3’ gcADPR-bound and 2’,3’-cGAMP-bound CbTad1 protein. 1”-3’ gcADPR and 2’,3’-cGAMP are shown as orange and yellow sticks, respectively. The two loops that undergo conformational changes upon ligand binding are highlighted. c, Detailed binding between CbTad1 and 2’,3’-cGAMP. Residues involved in 2’,3’-cGAMP binding are shown as sticks. Red dashed lines represent polar interactions. 2Fo-Fc electron density of 2’,3’-cGAMP within one binding pocket is shown and contoured at 1 σ. d, ITC assays to test binding of 2’,3’ –cGAMP to CbTad1 mutants. Representative binding curves and binding affinities are shown. The KD values are mean ± s.d. (n=3 independent experiments). Raw data for these curves are shown in Supplementary Fig. 3. e, Native PAGE showed the binding of CbTad1 and its mutants to 2’,3’-cGAMP. For gel source data, see Supplementary Fig. 1. f, Overall structure of CmTad1 hexamer complexed with cA3 and 2’,3’-cGAMP. cA3 and 2’,3’-cGAMP are shown as green and yellow sticks, respectively. Two views are shown. 2Fo-Fc electron density of cA3 and 2’,3’-cGAMP within CbTad1 hexamer contoured at 1 σ.

Extended Data Figure 4.. Binding of 1”-3’-gcADPR by CbTad1.

a, Overall structure of CbTad1 hexamer bound to 1”-3’ gcADPR, which is shown as orange sticks. b, Detailed binding between CbTad1 and 1”-3’ gcADPR. Residues involved in 1”-3’ gcADPR binding are shown as sticks. Red dashed lines represent polar interactions. 2Fo-Fc electron density of 1”-3’ gcADPR within one binding pocket is shown and contoured at 1 σ. c, Native PAGE showed the binding of CbTad1 mutants to cA₃ and 1”-2’ gcADPR. For gel source date, see Supplementary Fig. 1. d, ITC assays to test binding of cA₃ to CbTad1 mutant. Representative binding curves and binding affinities are shown. The K_D values are mean ± s.d. (n=3 independent experiments). Raw data for these curves are shown in Supplementary Fig. 3.

Extended Data Figure 5.. Mutations in independent nucleotide binding sites of SBS Tad1 disrupt specific inhibitory activities

a, Plaque assay to test the activity of SBS Tad1 against Thoeris immunity in vivo. F10 phage was spotted in 10-fold serial dilutions on a lawn of P. aeruginosa cells expressing Thoeris operon genes (PAO1:Tn7 Thoeris SIR2), or without Thoeris (PAO1:Tn7 empty), electroporated with pHERD30T plasmids carrying SBS tad1 (wild type or mutant gene) or empty vector. b, Plaque assay to test the activity of SBS Tad1 against CBASS II-A^GA immunity in vivo. PaMx41Δacb2 was spotted on a lawn of Pa011 cells with deletion of CBASS operon (Pa011ΔCBASS II-A^GA) or Pa011 wild type cells (Pa011 wt), electroporated with pHERD30T plasmids carrying SBS Tad1 (wild type or mutant gene) or empty vector. c, Plaque assay to test the activity of SBS Tad1 against CBASS III-C^AAA immunity in vivo. JBD67Δacb2 phage was spotted in 10-fold serial dilutions on a lawn of P. aeruginosa cells expressing Pa278 CBASS operon genes (PAO1:Tn7CBASS III-C^AAA), or without the system (PAO1:Tn7 empty), electroporated with pHERD30T plasmids carrying SBS tad1 (wild type or mutant gene) or empty vector.

Extended Data Figure 6.. PhiKZ gp184 is Tad1 sponge protein that binds CDN/gcADPR molecules.

a, Sequence alignment among phiKZ Tad1 homologs and SBS Tad1. For Tad1 proteins from phiKZ, PA7, phiPA3, Phabio phages the sequences without N-terminal domain (putative packaging domains) were used for alignment. CDN/gcADPR binding sites are shown with red frames, CTN binding sites are shown with yellow frames. Arrows indicate mutations in the binding sites. The enumeration of start and end amino acid positions is shown for phiKZ gp184. b, ColabFold prediction of phiKZ gp184 structure. The Tad1 domain is in gold, and the putative packaging domain is in blue. The protease cleavage site position H109 is indicated with a red arrow. c, ITC assays to test binding of 3’,3’-cGAMP, 1”,3’- and 1”,2’-gcADPR to phiKZ Tad1. Representative binding curves and binding affinities are shown. The K_D values are mean ± s.d. (n=3 independent experiments). Raw data for these curves are shown in Supplementary Fig. 9. d, Native PAGE showed the binding of phiKZ Tad1 to cyclic nucleotides and gcADPR molecules. For gel source date, see Supplementary Fig. 1. e, Plaque assay to test the activity of phiKZ gp184 against CBASS II-A^GA immunity in vivo. phiKZΔgp184 was spotted on a lawn of P. aeruginosa cells expressing Pa278 CBASS operon genes (PAO1:Tn7CBASS III-C^AAA), Pa011 CBASS operon genes (PAO1:Tn7CBASS II-A^GA), Thoeris SIR2 operon genes (PAO1:Tn7 Thoeris I) or without the system (PAO1:Tn7 empty). Cells were electroporated with pHERD30T plasmids carrying phiKZ gp184 (109–230 amino acids) or empty vector.

Extended Data Figure 7.. SPO1 Tad2 does not bind to cyclic dinucleotides.

a, Native PAGE showed the binding of SPO1 Tad2 to cyclic nucleotides and gcADPR molecules. For gel source date, see Supplementary Fig. 1. b, The ability of SPO1 Tad2 to bind and release 3’,3’-cGAMP when treated with proteinase K was analyzed by HPLC. 3’,3’-cGAMP was used as a control. The remaining nucleotides after incubation with SPO1 Tad2 was tested. c, Overlay of sensorgrams from surface plasmon resonance (SPR) experiments, used to determine kinetics of SPO1 Tad2 binding to cyclic dinucleotides. Data were fit with a model describing one-site binding for the ligands (black lines).

Extended Data Figure 8.. HgmTad2 binds to cGG and gcADPR molecules.

a, Profile of ion exchange chromatography of HgmTad2 using Resource Q column (1 mL, GE Healthcare). Proteins in peaks 1–3 are collected separately and marked as State 1–3. The proteins in three states were then subjected to native PAGE and SDS-PAGE, respectively. b, Native PAGE showed the binding of HgmTad2 in three states to 1”,2’ gcADPR. For gel source date of a and b, see Supplementary Fig. 1. c, SPR analysis of HgmTad2 binding to 2’,3’-cGAMP. The data was fitted with affinity model and the calculated K_D was shown.

Extended Data Figure 9.. The binding pockets of 1”-2’ gcADPR and cGG.

a, The binding pocket of 1”-2’ gcADPR in the HgmTad2–1”-2’ gcADPR structure. 2Fo-Fc electron density of 1”-2’ gcADPR is shown and contoured at 1 σ. b, Structural superposition among HgmTad2 in the apo form (two types of conformations) and 1”-2’ gcADPR-bound form. HgmTad2 in the apo form is colored orange and green for two types of conformations, respectively. HgmTad2 in the 1”-2’ gcADPR-bound form is colored cyan and pink for the two protomers. c, Native PAGE showed the binding of HgmTad2 and its mutants to 1”-2’ gcADPR. d, The binding pocket of 3’,3’-cGAMP in the HgmTad2–3’,3’-cGAMP structure. 2Fo-Fc electron density of 3’,3’-cGAMP is shown and contoured at 1 σ. e, Structural superposition between HgmTad2–3’,3’-cGAMP and HgmTad2-cGG. 3’,3’-cGAMP and cGG bind at the same position. f, The binding pocket of cGG in the HgmTad2-cGG structure. 2Fo-Fc electron density of cGG is shown and contoured at 1 σ. g, Structural superposition among HgmTad2 in the apo form (two types of conformations) and cGG-bound form. HgmTad2 in the apo form is colored orange and green for two types of conformations, respectively. HgmTad2 in the cGG-bound form is colored light magenta and pink for the two protomers. h, Closer view of the binding of the adenine base of 3’,3’-cGAMP in the binding pocket of HgmTad2. i, Native PAGE showed the binding of HgmTad2 mutants to 3’,3’-cGAMP. For gel source date of c and i, see Supplementary Fig. 1.

Extended Data Figure 10.. HgmTad2 has anti-Thoeris activity, but lacks anti-CBASS and anti-CRISPR-Cas activity in vivo.

a,b CapV enzyme activity in the presence of 3’,3’-cGAMP and resorufin butyrate. The enzyme activity rate was measured by the accumulation rate of fluorescence units (FUs) per second. HgmTad2 (8 μM) was incubated with 3’,3’-cGAMP (0.8 μM) for 10 min and then proteinase K (0.708 mg/mL) was added to release the nucleotide from the HgmTad2 protein. Filtered nucleotide products were used for the CapV activity assay. Data are mean ± SD (n=3 independent experiments). P-value: ****p< 0.0001. **p= 0.0012 c, Plaque assays with 10-fold dilutions of phage F10 to test the activity of SPO1Tad2 and HgmTad2 against SIR2 containing Thoeris system in vivo. Acb2/Tad2 expressed from p30T plasmid. d, Plaque assays with 10-fold dilutions of PaMx41Δacb2 to test the activity of SPO1Tad2 and HgmTad2 against Type II-A^GA CBASS in vivo. e, In vitro DNA cleavage with SpyCas9 (100nM), sgRNA (150 nM), substrate DNA (10 nM), and AcrIIA11/Tad2 proteins (10 μM). Cleavage produce presence indicates no inhibitor activity. For gel source date, see Supplementary Fig. 1. f, Plaque assays with 10-fold dilutions of phage JBD30 to test the activity of HgmTad2 against CRISPR-Cas9 system in P. aeruginosa.

Extended Data Figure 11.. GPBTad2 is active against Thoeris SIR2 and CBASS II-A^GA in vivo.

a, Plaque assays to test the activity of GPBTad2 against SIR2 containing Thoeris system in vivo. Organization of P. aeruginosa Pa231 Thoeris operon shown. F10 phage was spotted in 10-fold serial dilutions on a lawn of P. aeruginosa cells (PAO1) expressing Pa231 Thoeris operon genes (PAO1:Tn7 Thoeris SIR2), or cells without the system (PAO1:Tn7 empty), electroporated with pHERD30T plasmids carrying Acb2 and GPBTad2 genes or empty vector. b, Plaque assays to test the activity of GPBTad2 against Type II-A^GA CBASS in vivo. Organization of the P. aeruginosa Pa011 CBASS II-A^GA operon shown. PaMx41Δacb2 was spotted in 10-fold serial dilutions on a lawn of Pa011 cells with deletion of CBASS operon (Pa011ΔCBASS II-A) or Pa011 wild type cells (Pa011 wt), electroporated with pHERD30T plasmids carrying Acb2 and GPBTad2 genes or empty vector.

Figure 1.. Tad1 is a hexamer to bind to two molecules of cyclic trinucleotides.

a, ITC assays to test binding of cyclic oligonucleotides to CbTad1 and CmTad1. Representative binding curves and binding affinities are shown. The K_D values are mean ± s.d. (n=3 independent experiments). Raw data are shown in Supplementary Fig. 3. b, Ability of CbTad1 to bind and release cA₃ and 2’,3’-cGAMP when treated with proteinase K was analyzed by HPLC. c, Overall structure of CmTad1 hexamer. The Zn ion is shown as a sphere. Three views are shown. d, SLS assays of CbTad1 and CmTad1. Theoretical and measured values are shown in the figure. The chromatogram shows the elution profile of protein standards. e, Detailed binding in the hexamer interface of CmTad1. Residues involved in hexamer formation are shown as sticks. Red dashed lines represent polar interactions. f, SLS assays of CmTad1 and its mutant. 5× mut: CmTad1 ^{Q98A/E99A/M102A/W103A/K106A}. g, Overall structure of CmTad1 hexamer bound to cA₃.Two views are shown. Red circles mark the binding sites of gcADPR. h, Detailed binding between CmTad1 and cA₃. Residues involved in cA₃ binding are shown as sticks. Red dashed lines represent polar interactions. 2Fo-Fc electron density of cA₃ within one binding pocket is shown and contoured at 1 σ. i, Native PAGE showed the binding of CbTad1 and its mutants to cA₃ and 1”-2’ gcADPR. For gel source data, see Supplementary Fig. 1. j, ITC assays to test binding of 1”,2’-gcADPR to CbTad1 mutants. The K_D values are mean ± s.d. (n=3 independent experiments). Raw data are shown in Supplementary Fig. 3. k, Overall structure of CmTad1 hexamer bound to cA₃ and 1”-3’ gcADPR. cA₃ and 1”-3’ gcADPR are shown as green and orange sticks, respectively. 2Fo-Fc electron density of cA₃ and 1”-3’ gcADPR within CbTad1 hexamer contoured at 1 σ.

Figure 2.. Tad1 antagonizes Type II-A^GA and Type III-C^AAA CBASS immunity.

a, Isothermal titration calorimetry (ITC) of cyclic oligonucleotides binding to SBS Tad1 and ColiTad1. Representative binding curves and binding affinities are shown. The K_D values are mean ± S.D. (n=3 independent experiments). Raw data for these curves are shown in Supplementary Fig. 3. b, Summary of Tad1 binding results. X: no binding; W: weak K_D > 400 nM. S: strong K_D < 400 nM by ITC or SPR. c, CapV enzyme activity in the presence of 3’,3’-cGAMP and resorufin butyrate. The enzyme activity rate was measured by the accumulation rate of fluorescence units (FUs) per second. Data are mean ± S.D. (n=3 independent experiments). P-value: *****p< 0.0001. ns, not significant. d, CapV enzyme activity with Tad1 homologs. The experiment was performed as in c. For c and d, statistics were calculated in GraphPad Prism by applying the in-built analyses of one-way ANOVA with Dunnett’s multiple comparisons test vs. CapV only. P-value: *****p< 0.0001. ns, not significant. e, Phage plaque assays with F10 phage spotted in 10-fold serial dilutions against Thoeris or PaMx41Δacb2 against Type II-A^GA CBASS, with p30T plasmid expressing Tad1 genes. f, Agarose gel showing cAAA activation of NucC nuclease on DNA. N: nicked plasmid, SC: supercoiled plasmid, L: linear and cut denotes digested DNA. g, Plaque assays with JBD67Δacb2 phage spotted in 10-fold serial dilutions on a lawn of P. aeruginosa cells with or without the Pa278 CBASS operon. h, Plaque assays with WT JBD67 phage, JBD67 Δacb2 or JBD67 SBSTad1 (introduction of SBS tad1 gene in place of the acb2 gene in JBD67 genome), plated as in g. i, Plaque assays with phiKZ or phiKZΔorf184 phages was spotted in 10-fold serial dilutions on lawns of P. aeruginosa cells expressing the indicated defense system. For Figs. 3e–i, three replicates were performed for each experiment and the results were similar each time.

Figure 3.. Tad2 binds an array of cyclic dinucleotides.

a, The Fo-Fc density around the putative cGG in the structure of HgmTad2 of State 3 contoured at 2.5 σ. The density itself and with cGG placed are shown in the upper and lower panels, respectively. b, The molecules in HgmTad2 of three states released when treated with proteinase K was analyzed by HPLC. 3’,3’-cGAMP and cGG standard was used as controls. c, Native PAGE showed the binding of HgmTad2 of State 1 to cyclic oligonucleotides and gcADPR molecules. For gel source data, see Supplementary Fig. 1. d, Overlay of sensorgrams from surface plasmon resonance (SPR) experiments, used to determine kinetics of HgmTad2 binding to CDNs. Data were fit with a model describing one-site binding for the ligands (black lines). e, The ability of HgmTad2 of State 1 to bind and release 3’,3’-cGAMP when treated with proteinase K was analyzed by HPLC. 3’,3’-cGAMP standard was used as a control. The remaining nucleotides after incubation with HgmTad2 was tested. f, Overall structure of HgmTad2 tetramer. Two views are shown. g, Structure of a protomer of HgmTad2. Secondary structures are labelled.

Figure 4.. Tad2 binds cyclic dinucleotides and gcADPR molecules simultaneously.

a, Overall structure of HgmTad2 tetramer bound to 1”-2’ gcADPR, which is shown as gray sticks. b, Detailed binding between HgmTad2 and 1”-2’ gcADPR. Residues involved in ligand binding are shown as sticks. Red dashed lines represent polar interactions. c, ThsA enzyme activity in the presence of 1”-3’ gcADPR and ε-NAD. Wild-type (WT) and mutated HgmTad2 at 40 nM were incubated with 5 nM 1”-3’ gcADPR. And then the reactions were filtered and their ability to activate ThsA NADase activity was measured. Bars represent the mean of three experiments, with individual data points shown. Data are mean ± SD (n=3 independent experiments). Statistics were calculated in GraphPad Prism by applying the in-built analyses of one-way ANOVA with Dunnett’s multiple comparisons test vs. ThsA only. P-value: *****p< 0.0001. ns, not significant. d, Overall structure of HgmTad2 tetramer bound to cGG, which is shown as purple sticks. HgmTad2 is shown as surface model. e, Detailed binding between HgmTad2 and cGG. Residues involved in ligand binding are shown as sticks. Red dashed lines represent polar interactions. f, Native PAGE showed the binding of HgmTad2 mutants to cGG. g, Native PAGE showed the binding of HgmTad2 mutants to 1”-2’ gcADPR, 3’,3’-cGAMP or cGG. For gel source data of f and g, see Supplementary Fig. 1. h-i, Overall structure of HgmTad2 tetramer bound to cGG and 1”-2’ gcADPR simultaneously (h), or cGG and 1”-3’ gcADPR simultaneously (i), cGG, 1”-2’ gcADPR and 1”-3’ gcADPR are shown as purple, gray and orange sticks, respectively. 2Fo-Fc electron density of the ligands within HgmTad2 tetramer is contoured at 1 σ.

Figure 5.. Tad2 antagonizes Type I-D^GG CBASS immunity that uses cGG.

a, Structural superimposition between HgmTad2-cGG-1”-3’-gcADPR and SPO1 Tad2. HgmTad2 and small molecules are colored as in Fig. 5i. SPO1 Tad2 is colored gray. b, Sequence alignment among Tad2 homologs. Residues with 100 % identity, over 75 % identity and over 50 % identity are shaded in dark blue, pink and cyan, respectively. Secondary structural elements of HgmTad2 are shown above the alignment. The insertion region (residues 32–72) between β2 and β5 of HgmTad2 or between β2 and β3 of SPO1 Tad2 (residues 36–59) is marked with a rectangle. Biochemically studied Tad2 homologs are marked with an asterisk before its species name. c-d, SPR assay of SptTad2 (c) and SaTad2 (d). e, Summary of the binding results of Tad1 homologs. The figure is labelled as in Fig. 2b. f, Overall structure of SptTad2 bound to cGG. A close view of the bound cGG with 2Fo-Fc electron density contoured at 1 σ is shown in the lower panel. g, Structural superimposition between HgmTad2-cGG and SptTad2-cGG. HgmTad2 and cGG are colored as in Fig. 5i. is colored gray. SptTad2 and its bound cGG are colored gray. h, i, TIR-STING NAD⁺ cleavage activity in the presence of cGG and nicotinamide 1,N⁶-ethenoadenine dinucleotide (εNAD), which emits fluorescence when cleaved. The enzyme activity rate was measured by the accumulation rate of fluorescence units (FUs) per second. Data are mean ± SD (n=3 independent experiments). P-value: *****p< 0.0001. ns, not significant.