Abstract
Many bacterial species use Type VI secretion systems (T6SSs) to deliver anti-bacterial effector proteins into neighbouring bacterial cells, representing an important mechanism of inter-bacterial competition. Specific immunity proteins protect bacteria from the toxic action of their own effectors, whilst orphan immunity proteins without a cognate effector may provide protection against incoming effectors from non-self competitors. T6SS-dependent Rhs effectors contain a variable C-terminal toxin domain (CT), with the cognate immunity protein encoded immediately downstream of the effector. Here, we demonstrate that Rhs1 effectors from two strains of Serratia marcescens, the model strain Db10 and clinical isolate SJC1036, possess distinct CTs which both display NAD(P)+ glycohydrolase activity but belong to different subgroups of NADase from each other and other T6SS-associated NADases. Comparative structural analysis identifies conserved functions required for NADase activity and reveals that unrelated NADase immunity proteins utilise a common mechanism of effector inhibition. By replicating a natural recombination event, we show successful functional exchange of CTs and demonstrate that Db10 encodes an orphan immunity protein which provides protection against T6SS-delivered SJC1036 NADase. Our findings highlight the flexible use of Rhs effectors and orphan immunity proteins during inter-strain competition and the repeated adoption of NADase toxins as weapons against bacterial cells.
Similar content being viewed by others
Introduction
Bacteria typically exist in polymicrobial communities where they compete with closely- and distantly related microbes for space and resources, often by directly killing or disabling competitor cells. A widespread and important weapon used for competition between Gram-negative bacteria is the Type VI secretion system (T6SS), a large, contractile nanomachine used to deliver toxic effector proteins into neighbouring cells1. The T6SS functions by propelling a rigid puncturing structure decorated with effector proteins out of the secreting cell and into an adjacent recipient cell. This puncturing structure is composed of a tube of stacked rings of Hcp proteins, topped with a spike made from a VgrG trimer and a single PAAR (proline-alanine-arginine repeat) domain-containing protein. The expulsion of this structure from the secreting cell is driven by contraction of an extended cytoplasmic sheath-like structure anchored in a trans-membrane basal complex2,3. Effector proteins associate with components of the puncturing structure either through non-covalent interactions (cargo effectors) or by fusion of effector domains to core Hcp, VgrG or PAAR domains (specialised effectors)2. The T6SS can be used against eukaryotic cells, including host cells and fungal competitors, or to scavenge nutrients from the extracellular environment. However, the primary role of the T6SS appears to be during inter-bacterial competition, where it delivers multiple anti-bacterial effector proteins into a neighbouring competitor, causing death or inhibition of growth1,4. Many T6SS-delivered anti-bacterial effector proteins have been described, including families of peptidoglycan amidase and glycoside hydrolase effectors, phospholipases and pore-forming effectors, DNA hydrolases and deaminases, and effectors interfering with cellular cofactors2,5. In order to prevent intoxication of self or genetically-identical neighbour cells, secreting cells possess specific immunity proteins cognate to each anti-bacterial effector and encoded by the adjacent gene. These immunity proteins reside in the compartment of action of their effector protein and neutralise toxicity by tight and specific binding to the effector1.
The first T6SS effector reported to target an essential cellular cofactor was Tse6 from Pseudomonas aeruginosa, a PAAR-containing specialised effector whose C-terminal domain displays NAD(P)+ glycohydrolase (NADase) activity6. A subsequent study identified another NAD(P)+ glycohydrolase effector from Pseudomonas protegens, Tne2, and suggested that Tse6 and Tne2 represent founder members of two related families of NADase effectors, Tne1 and Tne2, respectively7. NADase toxins deplete available NAD(P)+, an abundant cofactor essential for cellular function in all kingdoms of life. Bacterial NADase toxins are also used against host cells, exemplified by the Mycobacterium tuberculosis tuberculosis necrotising toxin (TNT), an NADase domain cleaved from the CpnT (channel protein with necrosis-inducing toxin) protein. TNT induces macrophage cell death and is part of a distinct family of NADases found in many bacterial and fungal pathogens8.
The T6SS can be used for competition between closely related strains of the same species and there is considerable variation in effector-immunity genes within, as well as between, species1,9. Additionally, orphan immunity proteins lacking a cognate effector are believed to provide protection against effectors delivered by competitors10, although experimental evidence to support this idea is currently limited. Inter-strain variation and plasticity in T6SS effectors is exemplified by the rearrangement hotspot (Rhs) class of specialised effectors. Rhs proteins are large, polymorphic toxins with a highly variable C-terminal toxin domain (CT) preceded by a conserved Rhs-repeat containing domain which forms a shell-like structure around the CT11. T6SS-associated Rhs proteins have an N-terminal region containing a PAAR domain and structures important for chaperone binding and target cell entry12. Rhs CTs have a range of predicted or demonstrated activities, including varied DNase and RNase domains. The cognate immunity protein (RhsI) is always encoded immediately downstream of the rhs gene, allowing CT-I units to be exchanged through homologous recombination in the conserved rhs regions1. Whilst Rhs proteins containing Tne2-family NADase CTs have been predicted bioinformatically7, Rhs-associated NADase activity has not yet been demonstrated.
Serratia marcescens is an opportunistic bacterial pathogen which occupies diverse environmental niches and represents a significant cause of hospital-acquired infections13. The model strain S. marcescens Db10 has a single T6SS which displays anti-bacterial and anti-fungal activity and delivers at least ten effector proteins, including two Rhs proteins14,15. Previous work showed that Rhs1 is dependent on the VgrG2 spike protein and a specific chaperone, EagR1, for delivery and that its CT (Rhs1CTDb10) is a cytoplasmic-acting anti-bacterial toxin14,16. However, the mode of action of Rhs1CTDb10 has remained unclear. Rhs1CTDb10 is neutralised by the immunity protein, RhsI1Db10, encoded immediately downstream of rhs1, whilst the function of the proteins encoded by two other small genes immediately downstream of rhsI1Db10 at the 3’ end of the T6SS gene cluster is unknown. In this study, we use a combination of structural, biochemical and genetic approaches to show that Rhs1CTDb10 is an NAD(P)+ glycohydrolase toxin and is directly inhibited by RhsI1Db10, with RhsI1Db10 representing a previously undescribed family of immunity proteins. Additionally, we report that a clinical isolate of S. marcescens, strain SJC1036, has an Rhs1 protein whose CT (Rhs1CT1036) represents a distinct family of NAD(P)+ glycohydrolases, emphasizing the broad utility of such toxins in bacterial competition. Finally, by engineering Rhs1 from S. marcescens Db10 to deliver Rhs1CT1036, we demonstrate that one of the genes downstream of rhsI1Db10 in Db10 encodes an orphan immunity protein able to protect against Rhs1CT1036 and provide evidence for a patchwork of in-use and orphan immunity proteins in the Rhs1 locus in Serratia.
Results
The crystallographic structure of the Rhs1CTDb10-RhsI1Db10 effector-immunity complex
In preliminary work, remote protein homology and structural prediction suggested a possible distant relationship between Rhs1CTDb10 from S. marcescens Db10 and the CTs of Tse6 and Tne2, but there was insufficient similarity to assign function. Therefore, in order to elucidate the molecular function of this effector domain, we initiated a structural study. His6-tagged Rhs1CTDb10 (His6-Rhs1CTDb10, amino acids 1333-1473 of full length Rhs1) was co-expressed with its immunity protein, RhsI1Db10, and a stable 37 kDa heterodimeric His6-Rhs1CTDb10-RhsI1Db10 complex was isolated, demonstrating a direct physical interaction between the effector and immunity protein. The Rhs1CTDb10-RhsI1Db10 complex was purified to homogeneity, crystallised and the crystal structure determined to 1.3 Å resolution. The phases for the first electron density map were determined based on the resonance scattering of bromide ions that were soaked into the preformed crystals.
The Rhs1CTDb10-RhsI1Db10 complex is a globular entity with approximate dimensions of 50 × 45 x 60 Å. The structure places Rhs1CTDb10 as a member of the broad TNT-like family of bacterial NADases, which includes TNT, Tse6 and Tne2, displaying the palm domain characteristic of this enzyme family8. The palm domain is dominated by a complex, twisted β-sheet comprising seven strands in order 1-5-6-3-2-4-7, aligned in anti-parallel fashion (Fig. 1). There are six helices, the N-terminal α1, then α2 and α3 that link β3 with β4, then a short α4 is present between β5 and β6 creating a cavity, and finally α5 that precedes the C-terminal β7. The ~18 Å groove extending into the core of the protein along β5 and β6 represents the putative active site. This cavity is lined by residues on the turn between α1 and β1, strands β2 and β3 including the short turn linking these two elements of secondary structure, β6, and a loop between α3 and β4. The immunity protein RhsI1Db10 displays an α/β fold not observed previously, although in similar fashion to the effector domain, it is founded around a seven stranded β-sheet. The strands are anti-parallel in the order 4-5-6-7-10-9-8 (Fig. 1). On one side of the β-sheet, on the surface of the effector-immunity complex, lies the N-terminal segment of the immunity protein which adopts a three stranded antiparallel sheet, in order 1-2-3, then a loop including a short α1 that leads to β4. The other side of the main sheet, together with contributions from the loops linking β7 with β8, β9 with β10, and extending from β10 to α4, forms the interface to interact with and block the activity of Rhs1CTDb10 by occluding the putative catalytic cleft (Fig. 2a).
The protein:protein interface in the Rhs1CTDb10-RhsI1Db10 complex covers approximately 1440 Å2 and uses 15% of the solvent accessible surface area of Rhs1CTDb10 (9435 Å2) and 20% of RhsI1Db10 (7315 Å2). These values are indicative of a stable protein-protein complex17. The number of residues involved in interactions with the partner are 37 (about 28% of total residues) and 43 (about 26.5% of total residues) from Rhs1CTDb10 and RhsI1Db10, respectively. These residues form 16 direct hydrogen bonds and 10 salt bridges between the partners in addition to extensive van der Waals interactions. There are also numerous water-bridged hydrogen bonding interactions that will contribute to the association of the partners (Supplementary Fig. 1). The calculated free energy of complex formation, ΔG, is −13.6 kcal/mol, a value consistent with other T6SS effector-immunity pairs such as Tne2CT-Tni2 at −12.4 kcal/mol and Tse6CT-Tsi6 at −9.6 kcal/mol7 and these values are again consistent with the formation of a highly stable complex.
Of particular note is the RhsI1Db10 loop between β9-β10 that is directed deep into the Rhs1CTDb10 cleft (Fig. 2b). The tip of this inhibition loop forms a tight turn and presents four residues, Asp129-Pro130-His131-Asn132, that interact in a highly specific manner with residues in the Rhs1CTDb10 cleft to provide an effective block of the active site. Central to this is RhsI1Db10 His131, which participates in π-stacking interactions with Phe1386 of Rhs1CTDb10 on one side and van der Waals interactions with Tyr1359 and Val1409 on the other. The imidazole ND1 and NE3 form hydrogen bonds with solvent networks that bridge the partners. The side chain of Asp129 accepts a hydrogen bond from the His131 amide which helps to hold it in place to also accept a hydrogen bond from the hydroxyl group of Rhs1CTDb10 Tyr1359, serving to position this aromatic side chain to interact with the imidazole of His131. The RhsI1Db10 Pro130 abuts the three aromatic side chains of His131, Tyr1359 and Phe1386. The functional groups on Asn132 participate in solvent mediated links to the partner, in addition to a direct hydrogen bond between ND2 and the carbonyl of Leu1410. The positioning of RhsI1Db10 His131 in the cleft of Rhs1CTDb10 also provides insight into the catalytic mechanism of the effector (see below). The central role for His131 of RhsI1Db10 might suggest it is essential for Rhs1CTDb10-RhsI1Db10 interaction and toxin neutralisation. On the other hand, the nature of the interface between the two proteins, detailed above, implies that the interaction is unlikely to depend on any single residue and incorporates intrinsic robustness (belt and braces) to guarantee protection against the toxin. In support of the latter scenario, which we believe is likely to be common for effector-immunity interactions, mutation of His 131 to alanine did not prevent RhsI1Db10 from being able to neutralise Rhs1CTDb10 activity (Supplementary Figure 2).
Rhs1CTDb10 is an NAD(P)+ glycohydrolase
The most closely related proteins to Rhs1CTDb10 identified in the Protein Data Bank (PDB) were the two T6SS-dependent NAD(P)+ glycohydrolase toxins, Tne2CT from P. protegens Pf-5 (PDB 6B12; Z-score 11.1 and r.m.s.d. of 1.69 Å, over 101 and 95 equivalent Cα positions, respectively) and Tse6CT from P. aeruginosa PAO1 (PDB 4ZV0; Z-score 9.2 and r.m.s.d. of 2.48 Å, over 93 and 95 equivalent Cα positions, respectively)6,7. In addition, a well-characterised fungal surface NADase, AfNADase from A. fumigatus18, was identified that, although limited in overall similarity, provides important details related to enzyme mechanism (PDB 6YGF; Z-score 7.6 and r.m.s.d. of 2.18 Å, over 94 and 92 equivalent Cα positions, respectively). These observations suggested that Rhs1CTDb10 might be an NADase toxin.
In order to demonstrate NADase activity in vitro, we required purified Rhs1CTDb10. However, it proved impossible to produce Rhs1CTDb10 in the absence of RhsI1Db10 due to its toxicity towards the producing E. coli cells. Therefore, we adopted the approach of co-expressing and co-purifying His6-Rhs1CTDb10 with RhsI1Db10, followed by denaturation of the His6-Rhs1CTDb10-RhsI1Db10 complex with 8 M urea or 6 M guanidinium hydrochloride and subsequent on-column refolding of His6-Rhs1CTDb10 (described fully in Materials and Methods). In order to determine whether Rhs1CTDb10 has NADase activity, we incubated purified His6-Rhs1CTDb10 with the potential substrates NAD+, NADP+, NADH and NADPH, then separated the reaction products by HPLC and compared their retention times with standard compounds (Fig. 3). Incubation of His6-Rhs1CTDb10 with NAD+ resulted in hydrolysis of NAD+ to ADP-ribose and nicotinamide, and incubation with NADP+ also resulted in production of nicotinamide. No standard compound for pADP-ribose was available but a product peak slightly displaced from the NADP+ substrate peak is presumed to represent pADP-ribose in the latter reaction. In contrast, no activity of His6-Rhs1CTDb10 was observed against NADH or NADPH, with HPLC profiles following incubation of His6-Rhs1CTDb10 with these substrates being indistinguishable from no-protein controls (Fig. 3a). Confirming the specificity of the reaction, inclusion of RhsI1Db10, refolded separately from the His6-Rhs1CTDb10-RhsI1Db10 complex, prevented hydrolysis of NAD(P)+ (Fig. 3b, c). This direct neutralisation of Rhs1CTDb10 activity by RhsI1Db10 is consistent with its occlusion of the predicted active site of the toxin (Fig. 2). These data confirm that Rhs1CTDb10 is an NAD(P)+-glycohydrolase able to deplete the essential cellular cofactors NAD+ and NADP+.
The structure of Rhs1CTDb10 compared with other NAD(P)+ glycohydrolase toxins reveals clues about substrate recognition and mechanism
Having demonstrated that Rhs1CTDb10 is an NADase toxin, we compared its structure with those of the other NAD(P)+ glycohydrolase toxin domains, Tne2CT, Tse6CT and AfNADase (Fig. 4a). In all four proteins the palm domain is conserved, with β-strands particularly well aligned. Differences are primarily limited to the placement of helical segments on the periphery of the active site cleft. Since only the high-resolution structure of the Rhs1CTDb10-RhsI1Db10 complex was available, but not the effector domain in complex with appropriate ligands, we used structural and sequence comparisons with these other microbial NADases to help inform on aspects of effector specificity and mechanism.
The hydrolytic cleavage of the glycosidic bond between ADP-ribose and nicotinamide in NAD cofactors requires precise placement of the substrate to interact with a nucleophilic hydroxyl. During the catalytic process a cationic intermediate will be generated that supports C-N bond breakage and C-OH bond formation. Of interest with respect to the recognition of substrate by Rhs1CTDb10 is that the structure of AfNADase was obtained in complex with an NAD(P)+ mimic, benzamide adenine dinucleotide (PDB 6YGG)18. A structural overlay with the AfNADase ligand complex suggests that Rhs1CTDb10 presents potential phosphate interacting residues in Ser1384, His1412 and Arg1418, whilst Phe1386 is positioned to interact with the ribose moiety of NAD(P)+ (Fig. 4b).
The positioning of aromatic Tyr1359 and Phe1386 side chains might provide the electron rich environment able to support the development of a cationic catalytic intermediate (Fig. 4c). It is anticipated that a polar entity would assist the generation of a nucleophile by activating water. Deep on one side of the active site there are three residues, Gln1354, Asp1363 and Gln1452, of interest in this respect (Fig. 4c). The Asp1363 carboxylate, which is furthest away from the catalytic site, is fixed in position by hydrogen bonds accepted from the amide of Ile1438, and the amide and hydroxyl groups of Ser1439. In turn Asp1363 forms hydrogen bonds with Gln1354 and Gln1452 to position those side chains. An ordered water interacts with this glutamine pair, with distances of 2.8 Å and 3.5 Å respectively. This water forms a hydrogen bond, distance 3.1 Å, to His131 NE2 of RhsI1 (Fig. 4c). This polar feature within the cleft may generate the nucleophile or support interactions that correctly position the ribose of the substrate for catalysis to occur in Rhs1CTDb10.
Gln1408 and Gln1440 are positioned at the base of the cleft. The side chain of Gln1440 forms two hydrogen bonds with the amide and carbonyl of Gly1361. This positions Gln1440 NE2 to form hydrogen bonds to water that occupies the cleft, near the postulated ribose binding site. The side chain of Gln1408 is directed away from the cleft held down by a hydrogen bond between NE2 and the carbonyl of Gly1446. This positions Gln1408 OE1 to interact with a water that in turn interacts with Asp1363, and another water near the ribose binding site. An alternative rotamer of Gln1408 would position the side chain to stabilise the position of the ribose in the active site. Whilst the mechanism proposed previously18 is plausible, there remains a question of how the nucleophile is provided. Here, we have identified potential contributors to this aspect of the enzyme mechanism but further work will be required to elucidate it in detail.
The low sequence conservation and structural variation around the catalytic cleft of the four TNT-type NADases which we have compared suggest that whilst a similar reaction is catalysed, aspects of substrate recognition vary. There are however two notable features that are conserved: the placement of aromatic and hydrophobic side chains that appear to bind the reactive part of the substrate exemplified by Tyr1359, Phe1386 and Val1409 in Rhs1CTDb10, and a glutamine, Gln1452 in Rhs1CTDb10. In the latter case the residue is strictly conserved (Fig. 4d) and appears to be critical for placement of the nicotinamide ribose.
Investigation of residues potentially important for substrate binding and catalysis of Rhs1CTDb10
The structural model indicated which residues of Rhs1CTDb10 form the active site and that might contribute to NADase activity (Fig. 4). To investigate the potential contributions of selected residues for Rhs1CTDb10 activity, wild type Rhs1CTDb10 and mutant derivatives, each with a single candidate amino acid substituted, were produced in E. coli under the control of an arabinose-inducible promoter and the ability of each Rhs1CTDb10 variant to inhibit growth was assessed. For wild type Rhs1CTDb10, no growth was observed on induction with 0.02% or 0.2% l-arabinose. Three variants of Rhs1CTDb10 with amino acid substitutions of selected residues predicted to be important for function, namely F1386A, R1418A and Q1452A, showed complete loss of toxicity, indicating that these variants no longer possess NADase activity (Fig. 5a). Three variants of Rhs1CTDb10 with substitutions S1384A, S1399A and H1412A, showed a reduction in toxicity, with growth observed on 0.02% l-arabinose but not 0.2% l-arabinose. The structural model suggests that the changes to Ser1399 and His1412, residues distant from the active site cleft, have a minor effect probably due to destabilising the protein by removal of hydrogen bonding interactions. Ser1399 is located at the base of the active site and the side chain accepts a hydrogen bond from the amide of Phe1386, an important residue. Removal of this hydrogen bond might allow Phe1386 a greater conformational freedom leading to a reduction in enzyme efficiency. Such an effect would be expected given that a glycine is placed between the serine and phenylalanine. (Fig. 4c). In order to confirm that the Rhs1CTDb10 variants with reduced or no toxicity were still being produced, Rhs1CTDb10 was detected by immunoblotting using a 3xFLAG tag at the N-terminus of the protein. Non-toxic variants F1386A, R1418A and Q1452A were readily detected, confirming production of the protein (Fig. 5b). The variants displaying reduced toxicity, S1384A, S1399A and H1412A, were not detected, similar to wild type Rhs1CTDb10, consistent with even the partial activity of these variants being sufficient to disable the cell before high levels of protein can be produced. In the presence of RhsI1Db10, all variants were produced at similar levels to wild type Rhs1CTDb10 (Fig. 5b). Taken together, these data suggest that, as predicted from our structural analyses, amino acids Phe1386, Arg1418 and Gln1452 are essential for the activity of Rhs1CTDb10.
RhsI1Db10 is distinct from other NAD(P)+ glycohydrolase immunity proteins but displays commonalities in inhibition mechanism
The discovery that RhsI1Db10 has a previously undescribed fold confirmed that this protein is unrelated to the immunity proteins associated with Tse6CT, Tne2CT or TNT and, therefore, represents a distinct family of T6SS immunity determinants from those reported to date. As described above, the structure of the Rhs1CTDb10-RhsI1Db10 complex identified an inhibition loop in RhsI1Db10 which specifically and effectively blocks the active site of Rhs1CTDb10 with the key residue His131 (Fig. 2). Further examination of the structure of the complex revealed that the side chain of His131 from RhsI1Db10 occupies the site in the toxin where the substrate ribose would be placed. Phe1386 is predicted to form a π-π interaction with the ribose moiety of NAD(P)+ and RhsI1 appears to mimic this interaction as part of the complex formation with the effector by using the imidazole of His131. Although the immunity proteins that bind Rhs1CTDb10, Tne2CT and Tse6CT, namely Rhs1CTDb10, Tni2 and Tsi6, respectively, display distinct folds, they all accomplish enzyme inhibition by occluding the substrate binding site in a stable complex. It is particularly interesting to note that there is a common theme to these interactions. Each immunity protein places a basic residue at the site of catalysis. A structural overlay of the effector proteins, each in complex with their cognate immunity protein, indicates that His131 in RhsI1Db10 occupies a similar position to that of Lys62 in Tsi6 and Arg153 in Tni2 (Fig. 6). This similarity supports the idea that mimicking an aspect of the catalytic intermediate or transition state may contribute to the stability of the effector-immunity protein complex.
Rhs1 in a clinical strain of S. marcescens, SJC1036, contains a distinct NAD(P)+ glycohydrolase C-terminal domain
Whilst examining the genome sequences of a large collection of Serratia strains19, we noticed that a clinical isolate of S. marcescens, SJC1036, encodes an Rhs1 protein with a CT which might also represent an NADase toxin (Rhs1CT1036, amino acids 1333-1486). Sequence-based searches revealed that Rhs1CT1036 shares little similarity with Rhs1CTDb10 (21% sequence identity), but instead contains the conserved domain associated with TNT (pfam14021, E-value: 3.18 e−34). Encoded immediately downstream of Rhs1 in SJC1036 is a small, 124 amino acid protein predicted to be the corresponding immunity protein (RhsI11036). AlphaFold220 was used to generate a high confidence structural model of the predicted immunity protein, RhsI11036 (Fig. 7a, right). Inspection of this model in comparison with the available structure of RhsI1Db10 revealed that the two structures display different folds and have no apparent structural similarity.
To gain further insight into the function of Rhs1CT1036, a high confidence structural model of this effector domain was also generated using AlphaFold220 (Fig. 7a, left). Since Rhs1CT1036 was predicted to share similarity with TNT and the related NADase AfNADase, a structural comparison was performed (Fig. 7b, left). This revealed that the Rhs1CT1036 model was highly similar to AfNADase (6YGF; 34% sequence identity; Z-score 15.9 and r.m.s.d of 0.88 Å, over 107 and 103 equivalent Cα positions, respectively) and also similar to TNT (4QLP; 27% sequence identity; Z-score 10.3 and r.m.s.d of 1.70 Å, over 105 and 102 equivalent Cα positions, respectively). Furthermore, the alignment revealed that residues suggested to be required for binding NAD(P)+ overlapped across the three homologues (Arg1401, Phe1417, Gln1467 of Rhs1CT1036; Arg129, Phe137, Gln194 of AfNADase; and Arg757, Tyr765, Gln822 of TNT; Fig. 7c).
Comparing the structures of Rhs1CTDb10 and Rhs1CT1036 (r.m.s.d of 2.31 Å over 92 equivalent Cα positions), they appear to share a β-sheet core common to all NADases but are otherwise more divergent (Fig. 7b, right). Overall, the structural prediction provided strong evidence that Rhs1CT1036 was likely to be an NADase.
In order to determine whether Rhs1CT1036 does have NADase activity, we followed the same strategy as for Rhs1CTDb10. Recombinant His6-tagged Rhs1CT1036 was co-produced with RhsI11036, the complex isolated and His6-Rhs1CT1036 recovered alone by denaturation and refolding. Similar to Rhs1CTDb10, incubation of purified His6-Rhs1CT1036 with NAD+ resulted in its hydrolysis to ADP-ribose and nicotinamide, and incubation with NADP+ also resulted in production of nicotinamide, whilst no activity was observed against NADH or NADPH (Fig. 7d). Therefore, Rhs1CT1036 is also an NAD(P)+-glycohydrolase effector.
The S. marcescens Db10 Rhs1 locus encodes an orphan immunity protein which protects against Rhs1 from SJC1036
In S. marcescens Db10, there are two proteins encoded by genes immediately downstream of rhsI1Db10 at the 3’ end of the T6SS gene cluster whose function is unknown. However the fact that these genes are co-transcribed with the rhs1 and rhsI1Db10 genes21 implies a functional link with the T6SS. Comparison of the genomic sequences around rhs1 between Db10 and SJC1036 revealed that the first of these two genes of unknown function, SMDB11_2280, encodes a homologue of RhsI11036 from SJC1036 (Fig. 8a, b). Sequence alignment of the corresponding protein, 2280Db10, with RhsI11036 confirmed that the two proteins share 88% identity over their entire length and are almost identical after the first 12 amino acids (Fig. 8c). This suggested that 2280Db10 is an orphan immunity protein which might be able to provide protection against Rhs1CT1036 or related effectors delivered by competing strains. To test this hypothesis, we first determined whether 2280Db10 could provide protection against Rhs1CT1036 in an E. coli heterologous expression system. As expected, expression of Rhs1CT1036 alone resulted in severe inhibition of growth, and this toxicity could be fully alleviated by co-expression of RhsI11036. Strikingly, co-expression of 2280Db10 was also able to provide full protection against Rhs1CT1036, indistinguishable from that of the native immunity protein (Fig. 8d).
Next, we aimed to show that 2280Db10 can provide protection to Db10 against Rhs1CT1036 in the more relevant context of T6SS-mediated delivery. To do this, we engineered a strain of S. marcescens Db10 where rhs1CTDb10 and rhsI1Db10 (and the two downstream genes) have been precisely replaced with rhs1CT1036 and rhsI11036, named Db10 Rhs1_CTI1036 (Fig. 8b). In this strain, Rhs1CT1036-RhsI11036 are used in exactly the same way as the native CT-I pair, Rhs1CTDb10-RhsI1Db10 (and also the same as they would be in the original background, SJC1036). A control strain, Db10 Rhs1_no CTI, lacking Rhs1CT and all three downstream genes was also constructed (Fig. 8b). The Db10 Rhs1_CTI1036 strain (attacker) was co-cultured with target strains of Db10 encoding 2280Db10 (wild type Db10) or lacking 2280Db10 (Db10 Δ2280Db10), and the recovery of the target cells enumerated. If 2280Db10 provides protection against toxicity mediated by Rhs1CT1036, recovery of the Δ2280Db10 target will be reduced compared with that of the wild type parental strain, which should be fully resistant to Db10 Rhs1_CTI1036. We found that when Db10Δ2280Db10 was used as the target strain, there was a reduction in target cell survival when co-cultured with Db10 Rhs1_CTI1036 compared with when the attacker lacked Rhs1CT1036 (Db10 Rhs1_no CTI) or lacked a functional T6SS (ΔtssE Rhs1_CTI1036). Furthermore, when the engineered Rhs1 was introduced into a genetic background where every T6SS must incorporate and deliver Rhs1 (Δrhs2ΔvgrG1), inhibition of Δ2280Db10 was increased further (Fig. 8e). In contrast, there was no difference in the recovery of the parental strain encoding 2280Db10 whether co-cultured with attackers able to deliver Rhs1_CTI1036 or not. Therefore, 2280Db10 is indeed an orphan immunity protein able to provide Db10 with protection from competitors deploying Rhs1CT1036, whilst in the absence of 2280Db10, Db10 is susceptible to this TNT-like NADase effector. These data also confirm that, as expected, RhsI1Db10 (which is still present in Δ2280Db10) cannot protect against Rhs1_CT1036 (and, similarly, 2280Db10 cannot protect against Rhs1_CTDb10 since an ΔrhsI1Db10 immunity mutant is susceptible to Rhs1_CTDb1016).
We noticed that a homologue of RhsI11036/2280Db10 is also encoded downstream of an intact Rhs1-RhsI1 pair at the 3’ end of the T6SS gene cluster in Serratia ficaria 1D1416, suggesting that this is also an orphan immunity protein protecting against Rhs1CT1036-like NADases (Fig. 8a). Interestingly, the gene encoding the RhsI1 protein in this strain, which presumably protects against the preceding Rhs1CT of unknown function, appears to be homologous with SMDB11_2281, the second gene downstream of rhsI1 in Db10. This strongly suggests that SMDB11_2281 is also an orphan immunity protein, as well as showing that possession of an orphan immunity protecting against Rhs1CT1036-like NADase effectors is not unique to Db10.
Discussion
Rhs proteins are large polymorphic toxins that are widespread in Gram-negative bacteria and frequently associated with the T6SS. T6SS-associated Rhs proteins contain a highly variable C-terminal effector domain (CT), a shell-like structure which forms around the effector domain, and an N-terminal PAAR-containing domain with a functional role within the T6SS machinery11,22. This N-terminal domain includes a PAAR-domain which forms the tip of the expelled puncturing structure and often also transmembrane helices implicated in facilitating movement of the CT across the inner membrane of target cells12. Smaller, non-Rhs PAAR-containing effectors may have similar N-terminal PAAR and transmembrane helix-containing domains, which are bound and stabilised by related EagR-family chaperones in both cases, but they do not include the large shell domain of Rhs proteins12. Non-Rhs PAAR-containing effectors with NADase CTs have been reported previously, exemplified by Tse6 (renamed Tne1) and Tne26,7. In this study, we have shown that Rhs proteins can also incorporate CTs with NAD(P)+ glycohydrolase activity and have identified two distinct examples of such NADase CTs in one bacterial species.
Rhs1CTDb10 and Rhs1CT1036 are not closely related to Tne1/Tse6 or Tne2 effectors and appear to represent a distinctive group of Tne (Type VI secretion NADase effector). Whilst Rhs1CTDb10 is more closely related to Tse6CT and Tne2CT than other families of NADase toxins at a structural level, the sequence conservation is low (similar to that between Tne1/Tse6CT and Tne2CT), and several of the key amino acids predicted to be involved in substrate binding or catalysis are not identical between Rhs1CTDb10 and Tse6CT and/or Tne2CT. Additionally, the immunity proteins of Rhs1CTDb10, Tse6CT and Tne2CT are unrelated. On the other hand, Rhs1CT1036 is more closely related to TNT and AfNADase NADases, although again there are some differences in predicted key amino acids and the TNT immunity protein is not related to RhsI11036. Therefore, Rhs1CTDb10 and Rhs1CT1036 may represent additional sub-groups, Tne3 and Tne4, of T6SS-delivered NADase effectors. Despite the existence of distinct Tne groupings, it is interesting to note that all have very similar biochemical activity. Rhs1CTDb10 and Rhs1CT1036, just like Tne1/Tse6CT, Tne2CT, TNT and AfNADase6,7,18,23, are able to hydrolyse the oxidised cofactors NAD(P)+ but not the reduced forms NAD(P)H. The advantage of this pattern of specificity is not yet clear, but perhaps it represents an effective way to disrupt cellular NAD homeostasis, specifically the NAD(P)+/NADH ratio, and redox balance.
The existence of four distinct Tne groups, with likely more to be discovered, suggests that disruption of cellular NADase levels is an effective means to intoxicate a rival bacterial cell. This idea is further supported by the presence of Tne2-like toxin domains in putative effectors of the Type VII secretion system (T7SS) used for inter-bacterial competition in Gram-positive bacteria7, whilst the anti-bacterial T7SS of Streptococcus intermedius can also secrete a TNT-like NADase named TelB24. Tne2-like effector domains have also been identified in Rhs proteins7,25, although they have not been studied experimentally. NAD is an essential cellular cofactor involved in numerous and critical redox reactions and a steady level is critical for redox and energy homeostasis in the cell. NAD is one of the most abundant molecules in the cell, but whilst its total levels are high, much of it is bound within proteins and the amount of free NAD available is much lower, meaning that an enzyme able to cleave NAD(P)+ would be able to deplete cellular pools significantly, even if only one molecule is delivered at a time by the T6SS (as is the case for Rhs effectors)26,27. Effective depletion of available NAD(P)+ would then likely prevent the cell from being able to maintain its intracellular NAD(P)+/NADH ratio. Whilst this alone may not be enough to kill the targeted cell or produce the dramatic loss in viable target cell recovery observed with some nuclease effectors6,14,28,29, it is likely to have a synergistic effect with other effectors simultaneously delivered by the T6SS which damage cellular components or de-energise the cell. Indeed, it has been shown that Tse6 can be strongly synergistic with several other effectors in P. aeruginosa, including the membrane depolarising effector Tse4 and peptidoglycan hydrolase Tse130. Further supporting the idea that NADase enzymes may represent a widespread and effective strategy to disable bacterial cells, it has recently been reported that many different phage defence systems use unrelated NADases, containing sirtuin (SIR2) domains, to trigger a process known as abortive infection when infected bacterial cells die or arrest growth without producing phage progeny31.
Structural analysis of the Rhs1CTDb10-RhsI1Db10 complex, informed by comparisons with related systems, has advanced our knowledge of effector NADases and of how distinct immunity proteins protect against their potent destructive effect. Rhs1CTDb10 possesses a well ordered catalytic cleft, organised by a network of hydrogen bonding interactions and with a distinctive hydrophobic component. Of note, we have identified two key, conserved features of NADase activity that can be ascribed in Rhs1CTDb10 to residues Phe1386 and Gln1452. The aromatic residue is placed to support the attraction and binding of substrate, then repulsion of products, whilst the polar residue plays a key role in positioning of substrate for nucleophilic attack. For each NADase effector, the cognate immunity protein positions a basic amino acid at this conserved catalytic centre, mimicking the transition state that would exist during catalysis. Structures of the varied NADase effectors in complex with substrates, products and transition state-based inhibitors, married with detailed kinetic and thermodynamic studies, would, in the future, consolidate our understanding of substrate recognition and mechanism.
This study has also further highlighted the diversity of immunity proteins that are used to neutralise T6SS effector toxins. Whilst all four NADase effector domains have a broadly conserved fold and similar mechanism, the four cognate immunity proteins, RhsI1Db10, RhsI11036, Tsi6 and Tni2, display completely distinct folds. Indeed, in this study we show that RhsI1Db10 has a previously undescribed protein fold and RhsI11036 is predicted also to adopt a distinct, previously-unreported fold. Consistent with this, RhsI1Db10, Tsi6 and Tni2 all have distinct structures and modes of binding to their respective toxins. However, in a lovely example of convergent evolution, all three appear to neutralise the NADase toxin in a similar way. The common mode of inhibition by immunity proteins of NADase effectors relies on the formation of tight, high affinity complex, driven by extensive protein-protein interactions, and the placement of a highly structured loop harbouring a basic amino acid into the active site. This loop serves to occlude the active site from substrate entry whilst the basic residue appears to mimic an aspect of substrate binding and catalytic intermediate formation, namely the presence of a cationic intermediate or transition state. Such a feature may also contribute to the high affinity of the immunity proteins for their targets.
It is common to observe so-called orphan immunity proteins encoded immediately downstream of intact T6SS effector-immunity pairs1,9,10,32. These immunity proteins may be homologues of the in-use immunity protein but with a different effector specificity, or they may be from an unrelated family of immunity proteins. It has been widely assumed that such orphan immunity proteins provide protection against incoming effectors which the host strain does not possess, delivered by competitor cells of a different strain or species. However direct experimental evidence for this, particularly in the context of T6SS-mediated delivery, is limited. Here we demonstrate that 2280Db10 is a bone fide orphan immunity protein which is able to protect Db10 against the NADase effector Rhs1CT1036 from another strain of S. marcescens. 2280Db10 protects against Rhs1CT1036 as effectively as the native immunity protein RhsI11036 in an artificial expression system. Importantly, it also provides Db10 with full protection against T6SS-delivered Rhs1CT1036 at native expression levels. Thus, 2280Db10 is able to protect Db10 from intoxication by other strains of S. marcescens which carry a different CT on the same core Rhs protein. This functional demonstration of protection by an orphan immunity protein is consistent with observations by Ross et al. that members of the Bacteroidales possess mobile arrays of genes encoding orphan immunity proteins and that examples of such arrays could provide protection against two effector proteins delivered by the non-canonical Bacteroidales T6SS10. It is noteworthy that the presence of genes encoding standalone Rhs immunity proteins to protect against Rhs effectors delivered by other bacteria differs from the current paradigm for Rhs-encoding genetic loci. Previous work on other Rhs effectors, or analogous CdiA polymorphic toxins, has instead reported the accumulation of orphan CT-I (effector-immunity) modules33,34. In contrast, we report orphan immunity proteins without a corresponding CT encoded downstream of complete Rhs-RhsI pairs. This highlights an important additional aspect of immunity protein cross-protection and indicates a distinct pattern of CT-I acquisition and exchange in this class of Rhs protein. Interestingly, Rhs immunity proteins may also occur in isolation, as suggested by the recent identification of genes encoding putative Rhs immunity proteins in the absence of any rhs effector genes in Gilliamella35.
In this study we have demonstrated that Rhs proteins can be easily re-engineered to deliver and use new CT-I pairs, by replacing the Rhs1CT-I pair of Db10 with that from SJC1036 and demonstrating T6SS-dependent intoxication of parental Db10 by the newly engineered strain. It is believed that RhsCT-I pairs can be readily exchanged in nature via homologous recombination in the highly conserved core Rhs region, often leaving seemingly discarded CT-I pairs immediately downstream of an intact Rhs-RhsI unit16. Indeed, such an exchange resulting in a competitive advantage over the parental strain has been demonstrated in an experimental evolution setting33. This ability to readily exchange RhsCT-I pairs by homologous recombination helps to explain the huge diversity of Rhs CT-I pairs observed within, as well as between, species, and the corresponding importance of Rhs proteins in both intra-species and inter-species competition. Our engineering approach was designed to mimic such homologous recombination events by precisely replacing the CT-I unit but leaving the remainder of the Rhs1 protein intact (which is almost identical between the two strains). The success of this approach supports its future utility in demonstrating and testing the function of new Rhs CT-I pairs in a physiologically relevant, T6SS-delivery context rather than through their overexpression in isolation. We note the potential of this approach for the development of biocontrol or protein delivery strains designed to deliver specific effector or other protein domains in a biotechnological or therapeutic context.
Finally, we have provided further evidence for the diversity and genetic plasticity of T6SS effector-immunity pairs and orphan immunity genes, supporting the concept of a constant arms race between bacterial strains as they acquire the ability to kill (effectors) and to resist (immunity proteins) former siblings and new competitors. We have shown that the same toxic enzymatic activity (NADase) can be associated with the T6SS via at least four distinct effector domains. Furthermore, members of the same species can compete against each other using the same activity (NADase), deployed by the same basic effector chassis (Rhs1), through the use of distinct effector domains and immunity proteins. In the case of Db10 and 1036, the arms race has moved a step further as Db10 has acquired or retained an immunity protein (2280Db10) able to protect against the 1036-type NADase effector, as well as a further orphan immunity, SMDB11_2281, predicted to protect against an unrelated Rhs1 CT domain. Intriguingly a strain of S. ficaria which encodes an intact RhsCT-RhsI pair whose RhsI is homologous with SMDB11_2281, also encodes an orphan RhsI11036-like protein which likely protects against Rhs1CT1036-like NADase effectors. It seems clear that the arms race is dynamic and complex. It is tempting to speculate that Rhs1CTDb10-RhsI1Db10 was acquired in a horizontal gene transfer event replacing an Rhs1CT1036-RhsI11036-like unit and leaving 2280Db10 downstream to provide protection against the original strain. However, we do not have the genomic fossil record to ascertain the order of gene acquisition and replacement events in Db10 for certain.
In conclusion, this study has demonstrated the breadth and genetic plasticity of the repertoire of anti-bacterial NADase effectors delivered by the T6SS, as well as providing further insight into their activity and their neutralisation by diverse immunity proteins. We have shown that an orphan immunity protein linked with an Rhs NADase effector can provide full protection against a distinct Rhs NADase effector deployed by another strain during intra-species competition, in addition to demonstrating plug-and-play switching of effector domains between Rhs proteins. These findings support the concept of T6SS-mediated arms races between closely related strains, with the potential to shape a variety of polymicrobial communities.
Methods
Bacterial strains, plasmids and culture conditions
Strains and plasmids used in this study are detailed in Supplementary Table 1. Mutant strains of S. marcescens Db10 were generated by allelic exchange using the suicide vector pKNG10121 and streptomycin-resistant derivatives were generated by phage ΦIF3-mediated transduction of the resistance allele from S. marcescens Db1136. Derivatives of the pRSF Duet-1 and pET15b-TEV plasmids were generated for protein overexpression and purification, whilst plasmids for arabinose- and IPTG-inducible protein expression for toxicity assessment were derived from pBAD18-Kn and pET15b, respectively. Details of oligonucleotide primers or synthetic DNA fragments used in cloning are provided in Supplementary Table 2. Unless otherwise stated, bacterial cultures were grown in LB (LB (10 g L−1 tryptone, 5 g L−1 yeast extract, 10 g L−1 NaCl, with 1.2 g L−1 agar for solid media) at 37 °C for E. coli and 30 °C for S. marcescens. When required, media were supplemented with antibiotics: carbenicillin (Ap) 100 µg ml−1, kanamycin (Kn) 50 µg ml−1, streptomycin (Sm) 100 µg ml−1, chloramphenicol (Cm 25 µg ml−1); to maintain repression of proteins expressed from pBAD18-Kn, 0.5% glucose was added to the media for cloning and maintenance.
Recombinant protein production and purification for crystallography
The genes encoding the effector domain, Rhs1CTDb10, and the immunity protein RhsI1Db10 were cloned into the two multiple cloning sites of the expression vector pRSF Duet-1 to generate pSC962. This plasmid directs the expression of a stable heterodimeric complex with Rhs1CTDb10 carrying an N-terminal hexa-histidine tag (His6). A single colony of freshly transformed E. coli BL21(DE3) pLysS was used to inoculate 5 ml of LB media supplemented with Kn and Cm and cultured overnight at 30 °C. This culture was used to inoculate 1 L of the same media and incubated at 30 oC, shaking at 200 rpm, until an OD600 of 0.6–0.8 was reached. Expression was induced by addition of IPTG, to a final concentration of 500 μM, and the cultures incubated for a further 3 h. Cells were harvested by centrifugation (1500 g for 20 min) and washed in 50 mM Tris-HCl (pH 8). Cells were flash-frozen in liquid nitrogen and stored at −80 oC until required.
Cell pellets were thawed, resuspended in 10 ml lysis buffer (50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 20 mM imidazole and 5 mM β-mercaptoethanol) supplemented with complete EDTA-free protease inhibitor cocktail (Thermo Scientific) and DNase I (Sigma Aldrich), then passed through an Emulsi-Flex-C3 homogenizer (Avestin). Cell debris were removed by centrifugation at 40,000 g in a Beckman Avanti J-25, (JA 25.50 rotor) for 30 min at 4 °C and the supernatant further clarified through 0.2 µm filters prior to use in affinity chromatography. A 1 mL or 5 mL HisTrap HP (GE Healthcare) column charged with NiCl2 was pre-equilibrated with lysis buffer, and lysate containing His6-Rhs1CTDb10-RhsI1Db10 was loaded. The column was washed to remove all unbound proteins, and a five-step gradient of lysis buffer supplemented with 500 mM imidazole was applied. The complex eluted at approximately 250 mM imidazole. Fractions containing the complex were identified using SDS-PAGE, pooled and concentrated using a molecular weight cut-off spin concentrator (Millipore).
The protein complex was further purified using size-exclusion chromatography (SEC) on Superdex 75 HiLoad 16/600 column (GE Healthcare). This column was calibrated using molecular weight standards: blue dextran ( > 2,000 kDa), thyroglobulin (669 kDa), ferritin (440 kDa), aldolase (158 kDa), conalbumin (75 kDa), ovalbumin (43 kDa), carbonic anhydrase (29.5 kDa), ribonuclease A (13.7 kDa) and aprotinin (6.5 kDa) (GE Healthcare). The column was pre-equilibrated with two column volumes (CVs) of buffer (50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 1 mM TCEP) and proteins were loaded via 5 ml loop. The sample provided a well-defined profile with an elution volume of about 64.1 ml, corresponding to a molecular mass of ≈ 37 kDa. The theoretical molecular mass of the Rhs1CTDb10-RhsI1Db10 complex is 35.7 kDa. Fractions containing the complex were pooled and concentrated. SDS-PAGE confirmed the purity of the sample and the presence of two proteins with the mass of 17 and 19 kDa, corresponding to His6-Rhs1CTDb10 and RhsI1Db10, respectively. The yield of the complex was estimated as ≈ 2.5 mg L−1 of E. coli culture. Protein concentration was measured in a NanoDrop ND-1000 system (Thermo Scientific) using predicted molar extinction coefficient (Ɛ = 70,360 M−1 cm−1 at 280 nm) for the complex, obtained from ProtParam37. The purified protein complex was then used for crystallography.
Affinity purification and SEC was performed using an ÄKTA pure system equipped with Unicorn 6.4 software (GE Healthcare).
Crystallisation, X-ray data collection and processing
To identify the lead crystallisation condition, the protein complex was subjected to a range of commercially available screens in a 96-well sitting drop plate format. The first lead crystallisation condition was identified in the JCSG-plusTM screen (Molecular Dimensions) following a two day incubation at 20 oC in drop containing 300 nL of protein complex at 10 mg ml−1 and 300 nL of the reservoir solution containing 0.1 M Bis-Tris-HCl pH 5.5, 25% PEG 3350. These crystals were crushed to produce micro seed stocks for optimisation screening in a 24-well sitting drop plate format. Several serial dilutions of seed stocks were prepared, and seeds were introduced to crystallisation drops by passing a nylon loop through a seed stock solution and then dipping it into the crystallisation drops. Crystals with improved shape and dimensions were then observed following two days incubation in a condition containing 2 μl of protein at 10 mg ml−1 (50 mM Tris-HCl pH 8.0, 250 mM NaCl, 1 mM TCEP) and 2 μl of reservoir solution (0.1 M Bis-Tris pH 5.5, 25% PEG 3350).
Investigation of the diffraction properties of the crystals together with testing of cryo-protectants and soaking with bromide and iodide was carried out in-house. Crystals were passed through a solution of mother liquor adjusted to contain 250 mM NaBr, then dipped in PEG400 as a cryoprotectant and immersed and stored in liquid nitrogen. Data were collected at −173 oC, on beamline I03 at the Diamond Light Source (DLS, Didcot, UK) with a wavelength of 0.9150 Å, on the high energy side of the Br K-absorption edge, and an Eiger2 XE 16 M detector. Approximate anomalous scattering contributions at this wavelength are f’ −8.5 e- and f” 3.8 e-. The data were processed via the automated processing pipeline integrated with XDS38 and scaled in Aimless39.
Structure determination and refinement
The crystal displayed space group P1 with unit cell dimensions a = 39.66 Å, b = 44.11 Å, c = 46.81 Å, a = 101.20° b = 96.13° g = 114.15°. A Matthews coefficient of 2.01 Å3 Da−1 suggested a heterodimer in the asymmetric unit with solvent content of around 40% by volume. Initial phases were calculated using the Crank2 experimental phasing pipeline40, which located seven potential bromide atoms with over 25% probability cut off. Six of these were subsequently included in the refined model. The resulting electron density map at a resolution of 1.3 Å was of excellent quality and the first model was constructed using Buccaneer41. The Rwork and Rfree, the latter based on 5% of the data, were 0.3007 and 0.3276, respectively at this stage.
Rounds of electron and difference density map inspection, model manipulation in COOT42 and refinement in REFMAC543 led to a model consisting of residues Asn1342 to Leu1473 and Met1 to Tyr163 for Rhs1CTDb10 and RhsI1Db10 respectively. B-factors were refined anisotropically and hydrogen atoms were included. Water molecules were assigned to well-defined peaks in the difference density map ( > 3.5 s) that were within 2.5–3.5 Å distance from hydrogen bond donor and acceptor groups. Bromide ions, a molecule of Bis-Tris, and dual rotamers for several amino acid side chains were also included. MolProbity44 was used in combination with the validation tools provided in COOT to monitor model geometry during refinement. Crystallographic statistics are presented in Supplementary Table 3 and a portion of the electron density map is shown in Supplementary Figure 3. The coordinates and structure factors have been deposited with the Protein Data Bank under accession code 6XTD.
Structure prediction and in silico analysis
Structural and sequence comparisons were carried out using COOT42, Dali45 and XtalPred46; structural similarity values are given as Z score (Dali) and r.m.s.d. (generated by COOT v0.9.6). PDBePISA17 was used to analyse the surface interface between the effector and immunity protein. Molecular images were rendered using PyMOL v2.5.2 (Schrödinger). AlphaFold220,47 provided by Colab notebook (Google) was used to generate five prediction models from the query amino acid sequence and a model with the lowest PAE (Predicted aligned Error) and highest pLDDT (per-residue Local Distance Difference Test)48 scores was selected for subsequent analysis. Genomic synteny plot was generated using genoPlotR v0.8.11 and R v4.0.3.
Protein unfolding and refolding for isolation of individual components of RhsCT-RhsI complexes
Cleared cell lysates from cultures producing the His6-Rhs1CTDb10-RhsI1Db10 complex from pSC962, or the His6-Rhs1CT1036-RhsI11036 complex from pSC981, were generated and loaded on a 5 ml His Trap™ HP column as described above. The column was washed with 10 CV of lysis buffer (50 mM Tris-HCl pH 8.0, 250 mM NaCl, 20 mM imidazole, 5 mM β-mercaptoethanol) to remove unbound protein and then with 5 CV of 8 M urea to unfold the complexes and remove unbound immunity proteins. This was followed by a linear gradient from 100% 8 M urea to 100% buffer A over 10 CV. Bound proteins were then eluted with buffer B (50 mM Tris-HCl pH 8.0, 250 mM NaCl, 500 mM imidazole, 5 mM β-mercaptoethanol) and subjected to SEC using Superdex 75 HiLoad 16/600 column (equilibrated in 50 mM Tris-HCl, pH 8.0, 250 mM NaCl, 1 mM TCEP) to separate out the refolded toxin from any remaining effector-immunity complex that was not refolded. In the case of the experiment shown in Fig. 5b, unfolding was performed by incubating the protein complex in 5 M guanidinium hydrochloride at 90 oC for 5 min and then performing on-column refolding as described above. The immunity protein was recovered by dialysing the flow-through containing guanidinium hydrochloride against fresh buffer containing 25 mM Bis-Tris-HCl pH 6.0, 25 mM imidazole, 250 mM NaCl and 2 mM DTT.
NADase assays
Unless stated otherwise, reactions were prepared in 50 μl buffer (50 mM Tris-HCl pH 8.0, 250 mM NaCl) supplemented with 5 mM substrate (NAD+, NADH, NADP+, or NADPH), followed by addition of 500 μg purified Rhs1CTDb10 or Rhs11036 and incubation at 30 °C for 30 min. Then 50 μl acetonitrile was added to precipitate the protein and the precipitate was removed by filtration through a mini spin column (Neo Biotech). In the case of the experiment shown in Fig. 5b, reactions were performed in 100 μl buffer (50 mM Tris-HCl pH 7.5, 250 mM NaCl, 0.5 mM TCEP) containing 2.5 mM NAD+ (Rhs1CTDb10 + RhsI1Db10) or 5 mM NAD+ (Rhs1CTDb10 only), and either 30 μg RhsI1Db10 and 15 μg Rhs1CTDb10 (with RhsI1Db10 premixed with the substrate before the addition of Rhs1CTDb10), or 15 μg Rhs1CTDb10 only, were added, followed by incubation overnight at 30 °C. Analysis of standard compounds (5 mM) and reaction mixtures by HPLC was performed using the UltiMate 3000 HPLC system (Dionex) with Chromeleon (v6.8) software and an XBridge BEH-amide column (Waters), using a flow rate of 4 ml min−1. The column was equilibrated and washed between runs in 90% buffer A (95% v/v acetonitrile, 10 mM NH4CH3CO2, pH 8.0) and 10% buffer B (50% v/v acetonitrile, 10 mM NH4CH3CO2, pH 8.0). Compounds were separated using a linear gradient from 10% buffer B to 100% buffer B over 25 min (Figs. 3a, 7d) or 40 min (Fig. 3b) and absorbance was monitored at 280 nm.
In vivo toxicity assays
Genes encoding wild type 3xFLAG-Rhs1CTDb10 or variants with single amino acid substitutions were cloned under the control of an arabinose-inducible promoter in pBAD18-Kan and expressed in E. coli MG1655. Genes encoding Rhs1CT1036 with or without RhsI11036 and 2280Db10 were cloned under the control of an IPTG-inducible promoter in pET15b and expressed in E. coli BL21 (DE3) pLysS. Cells of freshly transformed E. coli grown overnight on solid media were resuspended in LB or M9 liquid media, normalised to OD600 of 1, serially diluted from 100 to 10−5, and 5 μl of each dilution spotted on LB or M9 agar plates with appropriate supplements. The plates were incubated overnight at 37 °C.
Protein levels of the 3xFLAG-Rhs1CTDb10 variants were visualised using an anti-FLAG immunoblot. Cultures of MG1655 carrying the above plasmids were grown overnight in LB + 0.2% glucose, subcultured 75 µl into 5 ml LB + 0.2% glucose and grown for 1.5 h, then subcultured 75 µl into 5 ml LB + 0.02% l-arabinose and grown for 2 h. Finally cells equivalent to 1 ml culture at OD600 1 were recovered by centrifugation, resuspended in 75 µl Laemmli SDS-PAGE sample buffer with β-mercaptoethanol, heated to 100 °C for 10 min. Finally, 5 µl of each sample was subjected to SDS-PAGE and immunoblotting using anti-FLAG primary antibody (Sigma, catalogue number F3165, 1:10,000) and HRP-conjugated anti-mouse secondary antibody (Bio-Rad, catalogue number 170-6516, 1:10,000). Uncropped and unprocessed blot images are supplied in the Source Data File.
Co-culture assays for T6SS-mediated anti-bacterial activity
Cells of relevant strains of S. marcescens Db10 grown overnight on solid LB media were resuspended in LB and normalised to OD600 0.5. The attacker and target were mixed at a 1:1 ratio and 25 µl of the mixture grown on solid LB at 30 °C for 7.5 h. Following the co-culture, cells were recovered in 1 ml LB and the number of surviving target cells was enumerated by serial dilution and viable counts on Sm-supplemented LB agar.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
The coordinates and structure factors generated in this study have been deposited in the Protein Data Bank under accession code 6XTD. All other data generated in this study are provided within the paper and its Supplementary Information files. Source Data are provided with this paper. Other structural data used in this study are available in the Protein Data Bank under accession codes 6B12 (Tne2CT-Tni2), 4ZV0 (Tse6CT-Tsi6), 6YGF (AfNADase), 6YGG (AfNADase in complex with benzamide adenine dinucleotide), and 4QLP (TNT). Bacterial genome sequences used in this study have Genbank accession codes GCA_000513215.1 (Serratia marcescens Db11), GCA_946406795.1 (Serratia marcescens SJC1036) and GCA_003641105.1 (Serratia ficaria 1D1416). Source data are provided with this paper.
References
Coulthurst, S. The Type VI secretion system: a versatile bacterial weapon. Microbiology 165, 503–515 (2019).
Jurenas, D. & Journet, L. Activity, delivery, and diversity of Type VI secretion effectors. Mol. Microbiol. 115, 383–394 (2021).
Wang, J., Brodmann, M. & Basler, M. Assembly and Subcellular Localization of Bacterial Type VI Secretion Systems. Annu Rev. Microbiol. 73, 621–638 (2019).
Gallegos-Monterrosa, R. & Coulthurst, S. J. The ecological impact of a bacterial weapon: microbial interactions and the Type VI secretion system. FEMS Microbiol. Rev. 45, fuab033 (2021).
de Moraes, M. H. et al. An interbacterial DNA deaminase toxin directly mutagenizes surviving target populations. eLife 10, e62967 (2021).
Whitney, J. C. et al. An interbacterial NAD(P)+ glycohydrolase toxin requires elongation factor Tu for delivery to target cells. Cell 163, 607–619 (2015).
Tang, J. Y., Bullen, N. P., Ahmad, S. & Whitney, J. C. Diverse NADase effector families mediate interbacterial antagonism via the type VI secretion system. J. Biol. Chem. 293, 1504–1514 (2018).
Sun, J. et al. The tuberculosis necrotizing toxin kills macrophages by hydrolyzing NAD. Nat. Struct. Mol. Biol. 22, 672–678 (2015).
Kirchberger, P. C., Unterweger, D., Provenzano, D., Pukatzki, S. & Boucher, Y. Sequential displacement of Type VI Secretion System effector genes leads to evolution of diverse immunity gene arrays in Vibrio cholerae. Sci. Rep. 7, 45133 (2017).
Ross, B. D. et al. Human gut bacteria contain acquired interbacterial defence systems. Nature 575, 224–228 (2019).
Jurenas, D. et al. Mounting, structure and autocleavage of a type VI secretion-associated Rhs polymorphic toxin. Nat. Commun. 12, 6998 (2021).
Ahmad, S. et al. Structural basis for effector transmembrane domain recognition by type VI secretion system chaperones. eLife 9, e62816 (2020).
Mahlen, S. D. Serratia infections: from military experiments to current practice. Clin. Microbiol. Rev. 24, 755–791 (2011).
Cianfanelli, F. R. et al. VgrG and PAAR Proteins Define Distinct Versions of a Functional Type VI Secretion System. PLoS Pathog. 12, e1005735 (2016).
Trunk, K. et al. The type VI secretion system deploys antifungal effectors against microbial competitors. Nat. Microbiol. 3, 920–931 (2018).
Alcoforado Diniz, J. & Coulthurst, S. J. Intraspecies Competition in Serratia marcescens Is Mediated by Type VI-Secreted Rhs Effectors and a Conserved Effector-Associated Accessory Protein. J. Bacteriol. 197, 2350–2360 (2015).
Krissinel, E. & Henrick, K. Inference of macromolecular assemblies from crystalline state. J. Mol. Biol. 372, 774–797 (2007).
Stromland, O. et al. Discovery of fungal surface NADases predominantly present in pathogenic species. Nat. Commun. 12, 1631 (2021).
Williams, D. J. et al. The genus Serratia revisited by genomics. Nat. Commun. 13, 5195 (2022).
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Murdoch, S. L. et al. The opportunistic pathogen Serratia marcescens utilizes type VI secretion to target bacterial competitors. J. Bacteriol. 193, 6057–6069 (2011).
Donato, S. L. et al. The beta-encapsulation cage of rearrangement hotspot (Rhs) effectors is required for type VI secretion. Proc. Natl Acad. Sci. USA 117, 33540–33548 (2020).
Tak, U. et al. The tuberculosis necrotizing toxin is an NAD+ and NADP+ glycohydrolase with distinct enzymatic properties. J. Biol. Chem. 294, 3024–3036 (2019).
Whitney, J. C. et al. A broadly distributed toxin family mediates contact-dependent antagonism between gram-positive bacteria. eLife 6, e26938 (2017).
Carobbi, A. et al. An antibacterial T6SS in Pantoea agglomerans pv. betae delivers a lysozyme-like effector to antagonize competitors. Environ. Microbiol. 24, 4787–4802 (2022).
Amjad, S. et al. Role of NAD+ in regulating cellular and metabolic signaling pathways. Mol. Metab. 49, 101195 (2021).
Canto, C., Menzies, K. J. & Auwerx, J. NAD+ Metabolism and the Control of Energy Homeostasis: A Balancing Act between Mitochondria and the Nucleus. Cell Metab. 22, 31–53 (2015).
Ma, J. et al. The Hcp proteins fused with diverse extended-toxin domains represent a novel pattern of antibacterial effectors in type VI secretion systems. Virulence 8, 1189–1202 (2017).
Salomon, D., Gonzalez, H., Updegraff, B. L. & Orth, K. Vibrio parahaemolyticus type VI secretion system 1 is activated in marine conditions to target bacteria, and is differentially regulated from system 2. PloS One 8, e61086 (2013).
LaCourse, K. D. et al. Conditional toxicity and synergy drive diversity among antibacterial effectors. Nat. Microbiol. 3, 440–446 (2018).
Garb, J. et al. Multiple phage resistance systems inhibit infection via SIR2-dependent NAD+ depletion. Nat. Microbiol. 7, 1849–1856 (2022).
Steele, M. I., Kwong, W. K., Whiteley, M. & Moran, N. A. Diversification of Type VI Secretion System Toxins Reveals Ancient Antagonism among Bee Gut Microbes. MBio 8, e01630–17 (2017).
Koskiniemi, S. et al. Selection of orphan Rhs toxin expression in evolved Salmonella enterica serovar Typhimurium. PLoS Genet 10, e1004255 (2014).
Poole, S. J. et al. Identification of functional toxin/immunity genes linked to contact-dependent growth inhibition (CDI) and rearrangement hotspot (Rhs) systems. PLoS Genet 7, e1002217 (2011).
Steele, M. I. & Moran, N. A. Evolution of Interbacterial Antagonism in Bee Gut Microbiota Reflects Host and Symbiont Diversification. mSystems 6, e00063–21 (2021).
Petty, N. K., Foulds, I. J., Pradel, E., Ewbank, J. J. & Salmond, G. P. A generalized transducing phage (phiIF3) for the genomically sequenced Serratia marcescens strain Db11: a tool for functional genomics of an opportunistic human pathogen. Microbiology 152, 1701–1708 (2006).
Gasteiger, E. et al. John M. Walker: Protein identification and analysis tools on the ExPASy server. Proteomics Protoc. Handb. https://doi.org/10.1385/1592598900.
Kabsch, W. Xds. Acta Crystallogr D. Biol. Crystallogr. 66, 125–132 (2010).
Evans, P. R. An introduction to data reduction: space-group determination, scaling and intensity statistics. Acta. Crystallogr D. Biol. Crystallogr. 67, 282–292 (2011).
Skubak, P. & Pannu, N. S. Automatic protein structure solution from weak X-ray data. Nat. Commun. 4, 2777 (2013).
Cowtan, K. The Buccaneer software for automated model building. 1. Tracing protein chains. Acta. Crystallogr. D. Biol. Crystallogr. 62, 1002–1011 (2006).
Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta. Crystallogr. D. Biol. Crystallogr. 60, 2126–2132 (2004).
Murshudov, G. N. et al. REFMAC5 for the refinement of macromolecular crystal structures. Acta. Crystallogr. D. Biol. Crystallogr. 67, 355–367 (2011).
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta. Crystallogr. D. Biol. Crystallogr. 66, 12–21 (2010).
Holm, L. & Rosenstrom, P. Dali server: conservation mapping in 3D. Nucleic Acids Res. 38, W545–W549 (2010).
Slabinski, L. et al. XtalPred: a web server for prediction of protein crystallizability. Bioinformatics 23, 3403–3405 (2007).
Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).
Mariani, V., Biasini, M., Barbato, A. & Schwede, T. lDDT: a local superposition-free score for comparing protein structures and models using distance difference tests. Bioinformatics 29, 2722–2728 (2013).
Acknowledgements
This work was supported by Wellcome (grant numbers 215599/Z/19/Z, S.J.C & W.N.H; 104556/Z/14/Z, Senior Research Fellowship S.J.C.; 220321/Z/20/Z, Senior Research Fellowship Renewal S.J.C.) and the Medical Research Council (grant number MR/N013735/1, PhD studentship M.H.). The Diamond Synchrotron Light source is acknowledged for beamtime. We would like to thank Kieron Lucas for assistance with HPLC and Paul Fyfe for crystallography support. We would also like to acknowledge Juliana Alcoforado Diniz for the construction of the Δ2280Db10 mutant, and Sarah Brough for assisting in constructing pSC3019 and pSC3023 and noting the homology between RhsI11036 and SMDB11_2280. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.
Author information
Authors and Affiliations
Contributions
M.H., W.N.H., and S.J.C. conceived the study; M.H., G.P., W.N.H., and S.J.C. designed experiments and analysed data; M.H., G.P., R.G.M., C.E., and G.B. performed experimental work; D.J.W. and G.P. performed bioinformatics analyses; G.P., W.N.H., and S.J.C wrote the manuscript with input from the other authors.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Hagan, M., Pankov, G., Gallegos-Monterrosa, R. et al. Rhs NADase effectors and their immunity proteins are exchangeable mediators of inter-bacterial competition in Serratia. Nat Commun 14, 6061 (2023). https://doi.org/10.1038/s41467-023-41751-3
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41467-023-41751-3
This article is cited by
-
Identification of type VI secretion system effector-immunity pairs using structural bioinformatics
Molecular Systems Biology (2024)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.