Modulation of Hachiman defence by a type II toxin-antitoxin system via balancing trade-off between the fitness cost and antiphage activity

Abstract

Hachiman systems provide innate antiphage immunity across prokaryotic domains. The system encodes a HamA nuclease and a HamB helicase both of which exhibit great diversity in sequence. Phylogenetic analyses of HamA and HamB proteins revealed similar phylogenetic trees for both proteins, falling into three major types. Close examination of one of the subclades identified a distinct subfamily in which most of these Hachiman systems stands alone, however, Hachiman in the Streptococcus genus is combined with PezAT, a distinct pneumococcal epsilon zeta toxin-antitoxin system, yielding the Pez-Ham system. Investigation of a S. thermophilus Pez-Ham system revealed that only the Hachiman system is required for mediating antiphage defence. Biochemical characterization of encoded proteins, i.e., HamA or HamB individually or in protein complex revealed that the HamA nuclease is inactive alone, but upon the formation of heterologous dimer with HamB, the resulting protein complex effectively cleaves DNAs of various forms with a broad specificity (5′-CNNNG-3′), and the nuclease activity is greatly facilitated by ATP-binding in HamB and to a less degree by ATP hydrolysis. Genetic investigations further showed, while the Pez system did not function in antiphage immunity in Escherichia coli, the system repressed the expression of Hachiman, and thereby balancing the trade-off between the fitness cost and the effectiveness of antiphage defence.

Graphical abstract

1. Introduction

Bacteriophages are viruses infecting prokaryotes, which outnumber their host cells by approximately 10-fold in most environments [[1], [2], [3]]. To defend against phage invasion, bacteria have evolved diverse antiphage defence systems, including the adaptive immune system CRISPR-Cas that eliminates nucleic acid invaders in the RNA-guided fashion [[4], [5], [6], [7], [8]], and restriction-modification (R-M) systems that target specific sequences in invaders for degradation [9] as well as numerous innate defence systems that induce abortive infection (Abi) to infected cells, thereby protecting uninfected cells in the population [[10], [11], [12]]. These antiviral systems often cluster together in their host genomes, forming “defence islands” [[13], [14], [15], [16]], and such genetic association has been used to predict new microbial immune systems for experimental demonstration [17,18]. To date, at least 145 new types of innate defence systems have been identified [19].

Among these new antiphage systems, those classified as Hachiman are present in 3.4 % prokaryotic genomes [17], carrying two genes encoding two core components of the defence system: HamA (containing a DUF1837 domain) and HamB (SF2-family helicase) [17]. To date, only a few Hachiman systems have been characterized, and these include (a) the minimal ones only carrying the two above-described domains, represented by the ECOR31 Hachiman complex [20], (b) the Escherichia coli anti-bacteriophage abpAB system in which the AbpA (or HamA) carries two essential nuclease domains, DUF1837 and Cap4 [21,22], and (c) the Sphingopyxis witflariensis DSM14551 Hachiman system with a 3-gene operon (hamC-hamA-hamB) [19]. Our analysis of Hachiman systems has revealed a distinct subgroup in which additional genes are associated with the two Hachiman core genes in certain bacteria, but not in others. Interestingly, the associated genes encode proteins highly similar to the toxin and antitoxin proteins encoded in the pneumococcal epsilon zeta toxin-antitoxin system (PezAT) in which PezT is a stable toxin that interferes bacterial cell wall synthesis [23] and PezA attenuates the toxicity by its antidote activity and this antitoxin forms a protein complex with PezT and functions as a transcriptional repressor to inhibit the expression of the toxin and antitoxin system [23,24]. Since their discovery, toxin-antitoxin systems have been primarily associated with plasmid stable inheritance and biofilms formation [25,26]. Recently, toxin-antitoxin systems were found to mediate antiphage defense [27,28] as well as to facilitate the stability of other phage defense systems such as CRISPR-Cas [29]. However, the co-localization of hamAB and pezAT genes in bacterial genomes suggests that these two systems could cooperate in anti-phage defense using a distinct mechanism.

Streptococcus thermophilus carries such a fused TA-Hachiman system denoted Pez-Ham. We found that the Hachiman system encodes a HamAB helicase-nuclease that cleaves DNA with broad specificity. Further characterization of the Pez-Ham system in E. coli showed that the PezAT system regulates the expression of the Pez-Ham operon while the Hachiman system is responsible for mediating antiphage defence, and the outcome of antiphage defence is determined by the expression level of the Hachiman system in bacterial cells.

2. Materials and methods

2.1. Bacterial and phage strains and incubation conditions

All bacterial and phage strains used in this study are listed in Table S1. Phage propagation was performed using E. coli BL21(DE3) as the host strain. The E. coli strain was inoculated in LB broth (10 g/L peptone, 5 g/L yeast extract, 10 g/L NaCl) and grown overnight at 37 °C. The resulting culture was then diluted in 1:100 ratio into fresh LB broth and further grown to an OD₆₀₀ = ca. 1. Then, 20 μL of phage stock was added and incubation was continued for 5–6 h or until cell lysis occurred. Cell debris was then removed by centrifugation at 5000 rpm for 10 min, and the supernatant was filtrated through a 0.22 μm membrane, yielding phage lysates/suspensions. The titers of phage in the lysates were estimated using plaque dot assay, which are typically 10¹⁰ to 10¹¹ PFU/mL.

For plaque formation assay, single colonies of E. coli strains harboring distinct plasmids were individually inoculated into 5 mL of LB broth and incubated at 37 °C until the OD₆₀₀ reached ∼1. Subsequently, 1 mL of the cultures was mixed with 9 mL of pre-warmed LB-agar (0.7 %) and poured onto plates. Phage stocks were serially diluted in LB. 2.5 µL of each dilution was spotted onto the solidified overlay. Plates were incubated overnight at 37 °C in an inverted position. The plaques on plates were recorded by taken pictures of them using the GE AMERSHAM ImageQuant 800.

2.2. Construction of phylogenetic trees of HamA and HamB proteins

Protein sequences associated with the Hachiman system were retrieved from the studies by Doron et al [17] and Payne et al [19], followed by preprocessing to remove duplicates and incomplete entries, giving an initial dataset of 1166 Hachiman systems from which representative sequences were selected with the 90 % similarity threshold using CD-HIT [30], and this yielded 193 HamA and HamB protein sequences that were analyzed by multiple sequence alignments using the online software MAFFT (https://mafft.cbrc.jp/alignment/software/). The phylogenetic tree was constructed using the Neighbor-Joining (NJ) method with 1000 bootstrap replicates to evaluate node reliability [31,32]. The resulting Newick-formatted tree was visualized and annotated by using the online software ChiPlot (https://www.chiplot.online/index.html) [33].

2.3. Construction of diverse plasmids

2.3.1. Construction of plasmids carrying defence systems and their mutated derivatives

Primers used in this study are listed in Table S2. PCR was conducted with Phanta Max Super-Fidelity DNA Polymerase (Vazyme) using a DNA template and a primer pair specified in each reaction, and the PCR program contained an initial denaturing at 95 °C for 5 min, 30 cycles of denaturing at 95 °C for 30 s, annealing at 55 °C for 30 s and extension at 72 °C for a time period of between 1 and 5 min, depending on the length of PCR products, and a final extension at 72 °C for 5 min. pBR322 was employed to clone defence systems for in vivo analysis of antiphage activity. The defence gene cluster of pezAT-hamAB was amplified from the genome of S. thermophilus ATCC19258 (GenBank ID: NZ_CP038020; coordinates 549,337–554,624) by PCR with the primer pair of PezATHamF and PezATHamR. The backbone of pBR322 was then amplified from the pBR322 plasmid using the primer pair of PBR322F and PBR322R. The two PCR fragments were assembled together at 50 °C for 30 min, using the MultiF Seamless Assembly Mix (ABclonal), and this yielded pPezHam. To construct pPez, the pezAT gene fragment was amplified from the bacterial genome by PCR using primers PezATF and PezATR, and the resulting gene fragment was assembled together with the pBR322 backbone fragment using the seamless assembly described above, giving pPez. pNpHam was also constructed in two steps: first, PCR fragments lacking the Pez system were generated by reverse PCR using HamorfdelF and HamorfdelR with pPezHam as a template; then, seamless assembly of the PCR fragments via the repetitive sequence designed on the two primers produced pNpHam carrying the fusion gene of the natural promoter and hamAB.

Plasmids carrying mutations in the active sites of HamA or HamB were generated by inverse PCR with primer pairs carrying individual mutation(s) (Table S2) using pNpHam plasmid as the template. 12 HamAB mutated derivatives were constructed (Table S1).

To construct pSynHam, a plasmid exhibiting strongly elevated level of expression for the Hachiman system, the hamAB genes was amplified from pPezHam by PCR using synHamF and synHamR, yielding a DNA fragment carrying a synthetic promoter with the native −35 box (TGGTGC) replaced by the −35 box (TTGACA) of a tetracycline resistance gene. The resulting gene fragments were assembled together with the pBR322 backbone fragment by the seamless assembly described above.

2.3.2. Construction of protein expression plasmids

To construct a vector for expressing proteins with multiple protein tags, the MBP gene fragment was amplified from pTac-His-MBP-PreScission (Beyotime Biotechnology) using MBPF and MBPR primers whereas the vector backbone fragments were amplified from pET30a (primers: pET-MBP-1 and pET-MBP-2). The two PCR fragments were then assembled together via seamless cloning, yielding pET-MBP that contains an N-terminal His-MBP tag and a C-terminal Strep tag. Then, pET-MBP fragment was amplified with the vector as the template using primers pETMBPF and pETMBPR whereas target genes (hamA and hamB) were obtained individually by PCR from pPezHam using primer pair of MBPHamAF and MBPHamAStrepR, MBPHamBF and MBPHamBR, respectively. Each gene fragment was assembled together with the pET30a fragment, yielding the expression plasmids of pETMBP-HamACStrep and pETMBP-HamBCStrep. Mutations in the active sites of hamA and hamB were generated by inverse PCR with primer pairs carrying individual mutation(s) (Table S2) using pETMBP-HamACStrep and pETMBP-HamBCStrep as the template, respectively.

2.3.3. Construction of toxicity detection plasmid

The pET-ara plasmid was constructed by amplifying the arabinose promoter element from the pBAD plasmid using araF and araR primers while the plasmid backbone was amplified from the pET30a plasmid template using primers pET-ara-1 and pET-ara-2. The pezA, pezT, pezAT, hamA, hamB, and hamAB genes were amplified via PCR using primer pairs listed in Table S2. Expression plasmids were obtained by assembling the vector and each of the gene fragment by recombineering following the seamless cloning procedure. The resulting constructs were transformed into E. coli BL21 (DE3). Plasmid-borne expression of each target gene was induced by adding l-arabinose to the culture medium to a final concentration of 1 %.

2.3.4. Construction of plasmids for protein expression level detection

The PCR amplification was performed using the pHamBHisF and pHamBHisR primer pairs (Table S2) with pPezHam, pNpHam, or pSynHamHis plasmids as templates. The amplified fragments were then ligated using the seamless cloning kit to obtain pPezHamHis, pNpHamHis, or pSynHamHis plasmids containing a 6xHis-tag at the C-terminus of the HamB protein.

2.4. Purification of HamAB proteins by recombinant protein expression in E. coli

Overnight cultures of E. coli strains harboring the protein expression plasmids were inoculated at 1 % (v/v) into 1 L of LB broth plus 50 µg/mL kanamycin and incubated at 37 °C for approximately 3 h until reaching an OD₆₀₀ of 0.6–0.7. After cooling to 12 °C, isopropyl β-d-1-thiogalactopyranoside (IPTG) was added to 0.5 mM to induce protein expression and the induction was for 16–18 h. Cell mass was harvested by centrifugation, and resuspended in Buffer A (20 mM Tris–HCl pH 8.0, 300 mM NaCl, 10 % glycerol, 20 mM imidazole). Cell lysis was conducted by sonication on ice using the SCIENTZ JY92-IIN carrying a No 6 amplitude transformer, with the setting of on: 3 s and off: 2 s for 60 min. Cell debris was removed by centrifugation at 10,000 rpm for 60 min at 4 °C. The resulting supernatant was loaded onto a Co-NTA column pre-equilibrated with Buffer A. After washing with 30 column volumes of Buffer A, the target protein was eluted with Buffer B (same as Buffer A but containing 300 mM imidazole). Eluted fractions were concentrated to ∼1 mL using a Millipore Amicon Ultra-15 centrifugal filter unit (MWCO 30,000), and the imidazole was removed by dialysis with Buffer A.

The MBP tag was cleaved overnight at 4 °C using tobacco etch virus (TEV) protease. For HamAB complex formation, tag-cleaved HamA and HamB proteins were mixed in 1:1 mass ratio and incubated overnight at 4 °C. The protein complex and HamB protein were further purified by size-exclusion chromatography using a Superdex 200 Increase column (GE Healthcare) equilibrated with Buffer C (20 mM Tris–HCl pH 8.0, 300 mM NaCl, 10 % glycerol). Eluted proteins were passed through a Ni-NTA column to further purify the protein to remove the uncleaved fraction. Final protein samples were concentrated and supplemented with glycerol to a final concentration of 50 % and placed at –20 °C for short term storage.

2.5. In vitro DNA cleavage assays

The oligonucleotide sequences used for preparing Fluorescent DNA Substrates are listed in Table S2. Fluorescent DNA substrates were generated by annealing 5′ or 3′-FAM-labeled single-stranded DNA oligos with equimolar amounts of their complementary unlabeled oligonucleotides in 5 × Annealing Buffer (Beyotime Biotechnology). The annealing reaction was performed using a BIO-RAD T100 Thermal Cycler by heating to 95 °C for 5 min, followed by gradient cooling to 25 °C in ca. 100 min.

DNA cleavage assays were performed in the cleavage buffer (20 mM HEPES, 5 mM Mg²⁺, 50 mM NaCl, 5 mM DTT). Different nucleic acid substrates individually specified in each experiment were incubated with HamA alone, or HamAB complex or its mutated derivatives at 37 °C for 30 min. To avoid the interference to gel electrophoresis by proteins in the reaction, Proteinase K was added to 1 mg/mL and treated in the Proteinase K buffer (10 mM EDTA, 1 % SDS, 50 mM HEPES) for 30 min. DNAs in the samples were then purified and analyzed by gel electrophoresis using an agarose or a polyacrylamide gel. FAM-labeled substrates were visualized using Ex/Em at 490/520 nm via a fluorescence channel, while non-fluorescent substrates were stained with EB and imaged using UV channels. Imaging was performed using the GE AMERSHAM ImageQuant 800.

2.6. Determination of the cleavage sites of the HamAB nuclease by sequencing cleavage products

DNA cleavage was set up with 200 nM HamAB and 100 ng pUC19 plasmid and incubated at 37 °C for 30 min, and then terminate with 50 mM EDTA. DNA cleavage products were separated by polyacrylamide gel electrophoresis according to their sizes, and DNA fragments of 100–250 bp were recovered by gel extraction. The cleavage fragments were processed, sequenced, with the sequencing data analyzed using the commercial sequencing service PE150 provided by GENEWIZ (Suzhou, China). Top 200 abundant sequences were identified from the sequencing results and aligned to the sequence of pUC19 plasmid. End sequences i.e., 6 bp in the upstream and the downstream of each aligned sequence and used for sequence enrichment analysis with the WebLogo program (https://analysis.genewiz.com.cn/).

2.7. ATP hydrolysis assay

ATPase activity of HamB, its mutants, and the HamAB complex was assessed using a colorimetric Malachite green assay to monitor free phosphate release [34]. The reaction was carried out in the a before mentioned cleavage buffer supplemented with ATP and protein at 37 °C. At designated time points, 10 µL samples were taken and quenched with 190 µL of 10 mM EDTA. Samples were then diluted on ice using 10 mM EDTA to generate a concentrations gradient for subsequent analysis. For phosphate detection, 200 µL of each diluted sample was mixed with 70 µL of Malachite Green Phosphate Detection Solutions [34,35], incubated at room temperature for 5 min, and absorbance at 630 nm was measured using a Feyond-A300 microplate reader (ALLSHENG). Use GraphPad software to plot enzyme kinetic curves based on the Michaelis-Menten equation [36].

2.8. DNA unwinding assay

To assess the helicase activity of HamB, 50 nM of 5′ FAM-labeled dsDNA substrates with varying lengths were incubated with HamB protein, its mutants, or the mutant HamAB complex at the indicated concentration gradients in cleavage buffer supplemented with 2 mM ATP. Additionally, 1 μM of unlabeled ssDNA complementary to the 5′ FAM-labeled strand was included to prevent reannealing. Reaction mixtures were incubated at 30 °C for 15 min and terminated by the addition of 50 mM EDTA. Samples were analyzed by 16 % PAGE under 100 V for 3 h at 4 °C. The fluorescent signal of FAM-labeled DNA was detected using the GE AMERSHAM ImageQuant 800.

2.9. Electrophoretic mobility shift assay (EMSA)

The binding of the HamAB complex or its mutants to FAM-labeled dsDNA substrate was evaluated by EMSA in a binding buffer containing 20 mM HEPES, 10 mM EDTA, 50 mM NaCl, and 5 mM DTT. The reaction was incubated at 30 °C for 30 min. Samples were resolved using 10 % native PAGE at 4 °C. FAM-labeled substrates were visualized using Ex/Em at 490/520 nm via a fluorescence channel of the GE AMERSHAM ImageQuant 800. The GraphPad software was used to plot enzyme kinetic curves based on the Hill equation [37].

2.10. Quantitative reverse transcription PCR (RT-qPCR)

RT-qPCR was employed to assess the transcriptional activity of the Hachiman system under different promoters. Strains harboring plasmids pPezHam, pNpHam, and pSynHam were cultured overnight and inoculated into fresh medium at a 1 % (v/v) ratio. Cultures were grown to an OD₆₀₀ of 0.6–1.0, after which 5 mL of each culture was harvested and washed with the PBS buffer. Total RNAs were extracted using the TRIgent reagent (Mei5 Biotechnology) by following the manufacturer’s protocol. 1 μg of total RNA per sample was reverse-transcribed using the Evo M-MLV RT Kit (Accurate Biology). qPCR was performed in a 10 µL reaction volume containing 1 µL of cDNA, 2 × SYBR Green Pro Taq HS Premix (Accurate Biology), the primer set for the Hachiman system or that for the gapA gene encoding a glyceraldehyde-3-phosphate dehydrogenase (listed in Table S2). Reaction was performed on the CFX96™ Real-Time PCR Detection System (Bio-Rad) with the following steps: denaturing at 95 °C for 30 s, followed by 39 cycles of 95 °C for 5 s and 60 °C for 30 s. Relative mRNA levels were evaluated using the comparative Ct (ΔΔCt) method, with gapA gene as the reference.

2.11. Western blot analysis

Strains containing plasmids pPezHamHis, pNpHamHis, or pSynHamHis were cultured in LB medium for 4–5 h until the OD₆₀₀ reached approximately 1–2. Then, 5 mL of bacterial culture was collected, centrifuged at 1200 rpm, and the medium was removed. The bacterial pellet was resuspended in 1 mL PBS, centrifuged again, and the supernatant was discarded to wash the cells. The pellet was then resuspended in 200 μL PBS and subjected to sonication for cell disruption. Cell lysis was conducted by sonication on ice using the SCIENTZ JY92-IIN carrying a No 3 amplitude transformer, with the setting of on: 3 s and off: 2 s for 2 min. Protein concentration was determined using the Detergent Compatible Bradford Protein Assay Kit (Beyotime Biotechnology). Protein samples (ca. 25 μg) were separated by SDS-PAGE. Western blot analysis was subsequently conducted following the method described by Liu et al [38]. The His antibody (Beyotime Biotechnology) was used at a dilution of 1:1000, and the reference antibody, purified anti-E. coli RNA polymerase α subunit (BioLegend), was used at a dilution of 1:5000. The HPR-labeled goat anti-mouse IgG (DINGGUO CHANGSHENG) was diluted at 1:15,000. Protein bands were visualized using the ECL Western blot substrate (Solarbio) and visualized by GE AMERSHAM ImageQuant 800.

2.12. Growth curve experiments

Single colonies of E. coli strains harboring either the Hachiman defence system or an empty vector (negative control) were picked from LB plates, inoculated in 5 mL LB broth and incubated at 37 °C until the OD₆₀₀ of the cultures reached ∼1.0. The cultures were then diluted with fresh LB medium to an OD₆₀₀ of ∼0.2, and the diluted cultures were then distributed into 48-well microplates with 700 µL/well. These aliquots of bacterial cells were infected with T2 phages at multiplicities of infection (MOIs) specified in each experiment. Growth curves were recorded for different treatments by measuring OD₆₀₀ every 10 min for 10 h using a Feyond-A300 microplate reader. Data were exported as CSV files and used for statistical analysis using the GraphPad Prism software. The infection experiments were performed in triplicates with different colonies of transformation.

3. Results

3.1. S. thermophilus encodes a distinct Hachiman system

Common to all identified Hachiman systems are the DUF1837 domain present in HamA and the SF2-family helicase domain in HamB. To gain an insight into their diversity in Hachiman systems, representative HamA and HamB proteins were retrieved from the identified Hachiman systems [17,19,20,22] and employed for constructing phylogenetic trees. We found that both HamA and HamB proteins form very similar trees, indicative of co-evolution of their encoding genes in Hachiman systems (Fig. 1a). Thus, Hachiman systems fell into at least three distinct types among which only the type I system present in E. coli K12 and a type III one in ECOR31 have been characterized [20,22], and they showed interesting differences in antiphage defence (see Discussion). Our analyses further revealed that type III Hachiman systems have two subgroups, denoted III-A and III-B, respectively. The ECOR31 Hachiman belongs to III-A, only consisting of hamA and hamB, whereas those of III-B either stand alone as for the III-A Hachiman or their Hachiman systems are clustered together with a PezAT toxin-antitoxin system or abiJ encoding a homolog of a protein implicated in abortive infection, constituting extended Hachiman systems (Fig. 1a, Fig. S1a, a). Interestingly, these types of extended Hachiman systems are exclusively present in the Streptococcus genus. In fact, the pezAT and hamAB systems could form an operon since the intergenic sequence between the last gene of pezAT and the first gene of hamAB is only 18 bp and the promoters of the pezAT-hamAB gene clusters are highly conserved in different Streptococcus species [24,39,40] (Fig. S2a). Thus, we were interested in testing whether and how these PezAT and Hachiman systems could function in antiphage defence in these bacteria.

Fig. 1.

S. thermophilus Hachiman system defends against phage invasion. (a) Phylogenetic tree of HamA and HamB proteins. The positions of the functionally characterized HamA or HamB proteins are highlighted in red on the trees, along with those encoded by the S. thermophilus Hachiman. The blue dots indicated the occurrence of the Pez-Ham systems. (b) Summary of plaque dot assay of the S. thermophilus Pez-Ham system with original data presented in Fig. S3a. Asterisk indicates a reduction in plaque size. (c) Conserved motifs in III-B HamA proteins enriched by MEME (https://meme-suite.org/meme/) analysis and mutagenesis of their highly conserved amino acids. (d) Conserved motifs in III-B HamB proteins were identified by multi-sequence alignments (shown in Fig. S4) [31] and mutagenesis of a few highly conserved amino acids. Annotations of these conserved motifs are as the following: Q (Q-loop); conserved motifs I-VI; WH (winged-helix domain); OB (oligonucleotide binding). Fold protection of HamB mutant was measured using series of 10-fold dilutions of phages.

To functionally characterize the two immune systems as well as their individual genes, attempts were made to construct expression plasmids for all of them by recombineering using linearized pETara, an arabinose-inducible expression vector, and PCR product of individual gene fragments. We found that while plasmids carrying the wild-type sequences were readily obtained for pETara-PezA, -PezAT, -HamA, -HamB, or -HamAB, very few colonies were obtained for pETara-PezT (Fig. S2b). Furthermore, sequencing the inserts in pETara-PezT colonies revealed that they carried indel mutations immediately after the start codon of pezT (Fig. S2c). These results indicated that the S. thermophilus PezT functions as a very potent toxin in E. coli and it kills bacterial cells upon expression. Since bacterial strains expressing pezAT gene pair were readily obtained (Fig. S2b), these results indicate that PezA effectively neutralizes the toxicity of PezT. To this end, we have demonstrated that PezAT also functions as a Type II TA system in E. coli. Strains carrying pETara-PezA, -PezAT, -HamA, -HamB, or -HamAB were further tested for their growth on glucose and l-arabinose LB plates, along with a reference strain carrying pETara, the cloning vector. We found they all grew equally well (Fig. S2d) on both plates, indicating that overexpression of each protein or each protein complex did not impair the cell growth.

Then, three expression plasmids were constructed using pBR322, including the whole pezAT-hamAB gene cluster(pPezHam), pezAT alone(pPez) and hamAB alone(pNpHam) and their expression was controlled by the original promoter (Fig. S2a). The resulting plasmids, pPezHam, pNpHam and pPez, were introduced into E. coli, and the resulting strains were employed to test the capability of each system to mediate antiphage defence, using 7 coliphages, including T1, T2, T3, T4, T5, T7 and Lambda. We found that the Hachiman system reduced the infection of T2 and T4 phages by up to 10⁶ folds when expressed directly from the promoter, but in the presence of pezAT, the defence strength was reduced by at least 100-fold (Fig. 1b; Fig. S3a). In addition, PezAT did not influence plaque formation (Fig. 1b; Fig. S3a).

S. thermophilus HamA and HamB were then characterized by mutagenesis of amino acids in their conserved motifs. HamA carried 3 conserved motifs in the DUF1837 nuclease domain, displaying slightly different consensus sequences across different types (Fig. S3b). AlphaFold prediction [41,42] showed that these conserved motifs formed a ββα-metal structural fold, which is a common active site for nuclease families [43] (Fig. S3c). Alanine substitution analysis of six highly conserved residues (G89, D90, E93, D127, E147, and K149) revealed that all but G89A mutations completely abrogated the Ham immunity against T2 infection in plaque dot assay (Fig. 1c, Fig. S3d), indicating that these HamA residues have essential roles in antiphage defence.

The HamB helicases contain 8 different motifs in the two Rec domains (Fig. 1d; Fig. S4a), consistent with their classification as members of SF2 helicase superfamily although the OB-fold domain only occurs in III-A HamB proteins among all known Hachiman subtypes (Fig.S4b). To characterize the III-B S. thermophilus HamB, 6 conserved residues (K275, R302, K354, D375, H580, T606) in the two Rec domains were chosen for alanine substitution mutagenesis, and this revealed that only the ATPase mutations (K275 in Motif I, D375 in Motif II) and T606 (Motif V) abolished the Ham immunity (Fig. 1d; Fig. S3d). Taken together, the S. thermophilus Ham system relies both on hamA and on hamB genes for mediating antiphage defence.

3.2. Nuclease activity of HamA is strongly activated by the cognate HamB ATPase

Structural similarity searches showed that the S. thermophilus HamA is highly similar to CT^TA271 (PDB:4G6U) (Fig. 2a), a Zn²⁺-dependent ββα-Metal DNase encoded by the CDI (Contact-dependent growth inhibition) system of E. coli EC869 [44,45]. To characterize the putative nuclease, HamAB complexes were reconstituted in vitro as described in Methods (Fig. S5a-c). The results were indicative of a heterologous dimer of HamAB, as reported for other Hachiman systems [20,22].

Fig. 2.

Biochemical characterization of the HamAB nuclease and its mutated derivatives. (a) Structural alignment of HamA with CT^TA271 (PDB: 4G6U). The regions with similar structures in the two proteins are displayed in dark blue (CT^TA271) and in orange (HamA), respectively, while the other regions are shown in gray or yellow. (b) Concentration-dependent cleavage of 100 ng supercoiled pUC19 DNA by HamAB complex. DNA fragments were separated on 1 % agarose gel and visualized with EB staining. OC, open circular plasmid; L, liner plasmid; CCC, covalently-closed circle DNA. (c) Determination of cleavage sites of HamAB complex. Left: DNA cleavage products were separated on 13 % PAGE. Those of 80–200 bp in size were recovered from the gel and sequenced by high-throughput DNA sequencing. Middle: end sequences of cleavage products were aligned using bioinformatic tools in the WebLogo website, generating enriched sequences. Right: the revealed cleavage site of HamAB. (d) Influence of HamA mutations (E93A, D127A) on the HamAB activity, and (e) Influence of HamB mutations (K275A, D375A) on the HamAB activity; 100 ng supercoiled pUC19 plasmid was incubated with a gradient of 40 nM-1 μM of the HamAB complex or its mutated derivatives as shown in the figure. Cleavage products were separated on 1 % agarose gel.

The purified HamAB complex, together with the MBP-tagged HamA and the TEV-treated HamA, was tested for nuclease activity using different substrates, including covalently-closed circle DNA (ccc-DNA), dsDNA, and M13 ssDNA and RNAs. We found that HamAB is an Mg²⁺/Mn²⁺-dependent nuclease (Fig. S5d) that cleaved DNAs of all tested forms into DNA fragments of ca. 100 bp (Fig. 2b; Fig. S5e), but ssRNA is not a substrate to this enzyme (Fig. S5f). In addition, neither the MBP-tagged HamA nor the TEV-treated one exhibited any detectable DNA cleavage activity (Fig. S5g, h). These results suggested that HamA is an inactive enzyme, and its nuclease has to be activated by HamB or by formation of protein complex with HamB.

Further characterization of the HamAB nuclease showed that the enzyme cleaved ccc-DNA to 20–100 bp fragments at a high concentration (1 µM) but producing linear DNA products at lower concentrations (e.g. 0.512pM) (Fig. 2b). Linear DNA fragments of ca. 80–200 bp generated with 0.2 µM HamAB (Fig. 2c) were recovered from polyacrylamide gels and sequenced by high-throughput sequencing. Analysis of the resulting data identified the 5′-CNNNG-3′ motif as the cleavage site for the HamAB complex (Fig. 2c).

To verify the site specificity, four DNA fragments carrying different numbers of 5′-CNNNG-3′, 5′-TNNNG-3′ and 5′-GNNNG-3′ sequences were employed as substrate to further test sequence preference of the enzyme, and we found that 5′-TNNNG-3′ motif also served as cleavage site for the S. thermophilus HamAB enzyme (Fig. S5i).

Among the HamA conserved residues, E93 and D127 are homologous to the acidic residues that coordinate metal ions in CT^TA271 (Fig. 2a). The importance of the conserved residues in HamA was tested by alanine substitutions, and this revealed the nuclease activity of the resulting HamA^E93AB and HamA^D127AB complexes was strongly impaired (Fig. 2d), suggesting that these HamA residues probably have the same role. However, testing the nuclease activity of the alanine substitution mutants in the ATP-binding and hydrolysis sites of HamB, i.e., HamAB^K275A and HamAB^D375A complexes, only moderately reduced the enzyme activity. In fact, inactivation of the ATP hydrolysis in HamAB^D375A showed a much weaker inhibition compared to HamAB^K275A, the mutant defective in ATP-binding (Fig. 2e). These observations raised an interesting question as to how HamB could influence the HamA nuclease in the effector complex.

3.3. DNA unwinding by HamB is facilitated by HamA, probably via DNA binding

HamB is a member of the SF2 family, carrying conserved motifs in both the ATPase and the helicase domains previously defined for SF1 and SF2 helicases [46]. Characterization of the purified recombinant S. thermophilus HamB either alone, or in complex with HamA showed that the HamB protein possesses strong ATP hydrolysis in the absence of DNA substrates (Km: 149±40.65 μM; Vmax: 27.76±2.145 μM/min) (Fig. 3a, b), and this is in contrast to conventional SF2 superfamily helicases for which the ATPase activity is DNA-substrate-dependent, including the two characterized HamB proteins [20,22]. Mutation of either the ATP binding site in HamB^K275A or the ATP hydrolytic site in HamB^D375A abrogated the ATP hydrolysis (Fig. 3c), verifying that these conserved motifs are responsible for the HamB ATPase activity. Moreover, the ATP hydrolytic activity of HamAB is very similar to that of HamB (Fig. 3d), indicating the formation of HamAB did not influence the ATPase configuration.

Fig. 3.

Functional characterization of the S. thermophilus HamB. (a) Effect of different DNA substrates on the ATP hydrolysis activity of HamB protein. Protein concentration 500 nM, ATP concentration 1 mM, DNA substrate addition: pUC19 plasmid, dsDNA, and M13ssDNA at 10ng/μL; 130 bp dsDNA, splayed dsDNA, 20 bp dsDNA, 20 nt ssDNA, 50 bp dsDNA, and 50 nt ssDNA at 50 nM. (b) Kinetic analysis of ATP hydrolysis by HamB. Km and Vmax values for ATP hydrolysis were determined as 149.8 ± 40.65 μM and 27.76 ± 2.145 μM/min, respectively. HamB protein: 500 nM. (c) Mutation of ATPase domains on the ATP hydrolysis activity of the wild-type HamB. HamB^K275A and Ham^D375A are ATP-binding and ATP hydrolysis mutants of HamB, respectively. HamB protein: 500 nM; ATP: 1 mM. (d) Comparison of ATPase activity between HamB and HamAB complex. HamB: 200 nM; ATP: 1 mM. (e) HamB Helicase assay. 50 nM of dsDNA of different lengths (20 bp, 30 bp, or 40 bp) was employed as substrate and assayed with the gradient (4, 20 or 100 nM) of HamB alone, or complexed a HamA mutant (HamA^E93A or HamA^D127A). (f) Quantitative analysis of the helicase activity of the 3 above enzymes. Reactions contained a concentration gradient of 12.5, 18.8, 37.5, 50, 75 or 100 nM of enzyme and 50 nM 20 bp dsDNA. All experiments were performed in triplicate; raw data of 3a-3c are provided in Table S3, and those of 3e in Table S4.

Testing the helicase activity of HamB showed the enzyme efficiently unwinds the 20 bp dsDNA substrates in an ATP-dependent manner, but is barely active against 30 or 40 bp dsDNA (Fig. 3e). Mutation of the ATP binding site (HamB^K275A) strongly inhibited DNA unwinding whereas mutation of the ATP hydrolytic residue (HamB^D375A) completely inactivated the enzyme (Fig. S6a). Two nuclease-dead HamA*B mutants (HamA^E93AB and HamA^D127AB) were then tested for DNA unwinding, and this revealed enhanced DNA unwinding both with the 20 bp dsDNA substrate as well as various substrates carrying the same 20 bp double-stranded region (Fig. S6b). But forming protein complexes with HamA did not change the preference of dsDNA sizes of the HamB helicase (Fig. 3e; Fig. S6b). Nevertheless, enzyme gradient assay (12.5 to 100 nM) showed that HamA*B was much more active than HamB alone since up to 10-fold difference in DNA unwinding was observed at low concentrations (Fig. 3f).

To test if HamA could facilitate the HamB helicase activity via DNA binding, EMSA was conducted for the wild-type HamAB, HamA^E93AB, HamA^D127AB as well as HamB alone. We found that the affinity of the HamAB complex for DNA substrates gradually increases with the length of the substrate (Fig. S6c; Table S5). Furthermore, while the HamB protein only showed weak DNA binding in the EMSA assay, HamA*B complexes carrying mutation in the conserved amino acids in the nuclease domain exhibited much higher DNA-binding affinity (Fig. S6d), thus providing a possible mechanism for the enhanced unwinding of 20 bp dsDNA by the HamA*B complexes.

3.4. Outcome of the Hachiman antiphage defence is controlled by the expression level of the antiphage system

Among all identified III-B Hachiman systems, only a small portion is associated with PezAT (Fig. 1a). Since PezAT regulates its own expression [24], we were interested to test how the Pez system could affect the expression of Hachiman in the same operon. Two different promoters were chosen with which three expression plasmids of hamAB were constructed, including SynHam in which the hamAB expression was driven by a synthetic promoter with a highly efficient E. coli −35 box element (Fig. S7a), and NpHam in which the gene expression was conferred from the native promoter as well as PezHam where gene expression could be regulated by the Pez system from the native promoter.

These expression plasmids were constructed and introduced into E. coli cells, yielding strains of SynHam, NpHam and PezHam individually. Then, total RNAs were prepared from cultures of these strains and analyzed for the relative expression levels of hamA and hamB genes by quantitative PCR. We found that both genes were expressed to different levels in these strains. In the PezHam strain, the expression of the four gene unit of pezTA-hamAB was the lowest, and an increase of ∼5-fold was found for hamAB where the gene transcription was directly conferred from the native promoter (Fig. 4a), and this is consistent with the previous finding that PezTA autoregulates its own expression [24]. Furthermore, expression of hamAB in the SynHam strain was conferred from the synthetic promoter in which an optimal −35 box was used to replace the original one, yielding a further increase of 3–4 folds expression (Fig. 4a). Subsequently, by introducing a His-tag at the C-terminus of HamB, we quantified the expression levels of Ham system proteins under different promoter strengths via Western blot analysis. The results demonstrated that in the PezHam strain (containing the PezAT system), where transcription of the Ham system was strongly suppressed, protein expression was consequently inhibited (Fig. 4b). After removing the PezAT system, the expression of Ham proteins in the NpHam strain increased by approximately 50-fold (Fig. 4b). In contrast, the protein expression of the Ham system in strains containing the SynHam plasmid was about 2-fold higher than that in the NpHam strain (Fig. 4b).

Fig. 4.

Influence of the Ham expression levels on the outcome of antiphage defence. (a) qPCR analysis of defence gene expression under different promoters. The gapA gene encoding a glyceraldehyde-3-phosphate dehydrogenase was used as the reference. Two primer pairs were designed (shown on the right), and used for quantitative PCR to estimate the expression of the hamAB genes. Error bars represent SD obtained by measuring each sample in five biological replicates. Raw data are provided in Table S6. (b) Protein expression levels of the Ham system under different promoter strengths. The reference protein is the E. coli polymerase α subunit with a molecular weight of approximately 40 kDa. Error bars represent SD obtained by measuring each sample in three biological replicates. Two-sided Student’s t-test was used for statistical analysis. P values are indicated above the plots. P < 0.05 was considered statistically significant. Raw data are provided in Table S7. (c) Plaque dot assays of antiphage defence by E. coli strains with differential expression of hamAB. 2 μl of phage dilutions was used in each dot assay. (d-g) Growth curves of E. coli strains with differential expression of the Ham system following T2 phage infection at different multiplicities of infection (MOIs).

These hamAB constructs were then employed to investigate the influence of hamAB expression on the antiphage activity both on plates and in liquid cultures. In plaque dot assay, the PezHam strain showed moderate resistance to phage infection, reducing plaque formation of T2 by ca. 10² folds and that of T4 by ca. 10³ folds, and a 10-fold more reduction of plaque formation was observed for the NpHam cells whereas an addition 10-fold reduction was obtained with the SynHam strain (Fig. 4c, Fig. S7b). These results are in full agreement with the correlation between the expression levels of antiphage effectors and the extents of antiphage defence, and consequently inhibition to their host cells.

The four strains were then tested to assess the influence of expression levels of Hachiman on the growth and antiphage activity in liquid cultures, and this revealed that, in the absence of phage, SynHam showed retarded growth probably reflecting the toxicity of the Hachiman effector when present at a high level whereas the remaining 3 strains (pBR322, pPezHam, pNpHam) grew similarly (Fig. 4d). The antiphage activity was tested at different MOIs, and we found that at an MOI of 0.05, the growth of the reference strain was strongly inhibited at ca. 1 h after infection and cell density started to drop at ca. 3 h, indicative of cell lysis whereas PezHam showed strong resistance to T2 infection in which the hamAB genes were expressed to the lowest level (Fig. 4e). However, the same level of the hamAB expression failed to protect host cells at MOI 0.5 although NpHam and SynHam showed strong defence activity against the phage (Fig. 4f). Strikingly, SynHam showed strong protection against phage invasion at MOI of 5 (Fig. 4g). Together, these results indicated that there is a trade-off between the expression level of the Hachiman system and the strength of antiphage defence in bacteria: expression of the antiphage system is harmful to bacteria on one hand but it increases the capability of antiphage defence on the other hand. Thus, this fused Pez-Ham systems represent a naturally occurring trade-off example, and further characterization will reveal how the system could impact the physiology of the bacterial host.

4. Discussion

Bacteria and archaea carry numerous defence systems that are often clustered together in defence islands in their genomes, suggesting that they may cooperate in mediating antiviral defence. Although these immune systems are highly dynamic in bacterial genomes as they are subjected to frequent gain and loss during evolution in a given organism, these systems are present in the pan-genome, comprising the sum of all immune systems available for a microorganism to horizontally acquire and use. Indeed, different antiviral systems are found to cooperate to expand the range of antiphage defence in bacteria [47]. However, how these defence systems may directly interact with each other remains elusive.

Herein, Pez-Ham, a fused PezAT and III-B Hachiman system of S. thermophilus has been characterized in E. coli, a heterologous host. We found that PezAT functions as a regulator to control the expression of the fused operon including pezAT and hamAB genes. We further demonstrate that the expression levels of Hachiman are positively related to the effectiveness of antiphage defence as shown in Fig. 4d. Recently, Aframian et al. reported that variations in expression levels can alter the phage resistance spectrum while also influencing self-inflicted toxicity [48]. Consistent with this finding, growth inhibition was observed in strains overexpressing the Hachiman system from a strong promoter in this work (Fig. 4d). As a result, all bacteria and archaea have to deal with the trade-off between the fitness cost and protection effectiveness. Currently, there are no relevant studies exploring which factors can influence the expression levels of the defense system. In S. thermophilus, the Hachiman defense system is genetically coupled with the type II toxin-antitoxin system PezAT within the same operon under a shared promoter. Type II antitoxins typically bind to promoter regions to suppress transcription [28]. Under environmental stresses such as phage infection, antitoxins are selectively cleaved by intracellular ATP-dependent proteases (e.g., Lon and ClpX) [[49], [50], [51], [52], [53], [54], [55]]. Thus, it is likely that PezA degradation relieves promoter repression, thereby activating the expression of downstream genes including hamA and hamB. Together, this suggests that S. thermophilus has evolved the Pez-Ham system to finetune the expression level of the Hachiman defence by coupling it to a PezAT system to balance the trade-off, representing the first such integrated defence system that has been characterized thus far.

The correlation between the expression level of defence system and the outcome of antiviral defence has already been observed for the Lactobacillus delbrueckii subsp. bulgaricus III-A Csm (LdCsm) system. It has been demonstrated that the LdCsm DNase facilitates abortive infection when the copy number of an invader is high [56], but mediating invader clearance when the invading nucleic acid is in a low abundance in infected cells [57]. The LdCsm system is a unique III-A CRISPR system since it solely relies on the Histidine-Aspartate (HD) nuclease for mediating immune responses [58] although this observation cannot be generally applied to other type III CRISPR-Cas systems which mainly rely on the cOAs signaling pathway for executing antiviral defence via abortive infection [59,60]. On the other hand, investigations of induced expression of several defence systems have revealed the trade-offs between the protective range and the cost of autoimmunity [48]. Clearly, carefully controlling the level of expression is not only important for the effectiveness of antiphage defence but also for the defence range, thus representing the aspect of defence physiology which has been underexamined to date.

Hachiman defence systems are widespread in bacteria and archaea and they fall into three main classes based on the sequences of HamA and HamB proteins: Type I, II and III. To date, several Hachiman systems have been tested for mediating antiphage defence among which two have been characterized in detail [20,22]. These studies have revealed interesting diversity in their possible mechanisms of antiphage defence and its activation. Investigation of Hachiman systems of B. cereus and E. coli K12 showed that activation of Hachiman systems can be associated with DNA damage repair, including co-expression of phage SSB protein, leading to growth inhibition or cell death [61,62]. Characterization of the Hachiman system of E. coli ECOR31 reveals that this Hachiman system can detect genome integrity through the HamB protein that recognizes the 3′-OH ends of broken DNA fragments. Thus, the Hachiman system functions in DNA damage sensing, in addition to the function in antiphage defence by facilitating intracellular nucleic acid degradation and eventually abortive infection [20]. Common for these observations is that Hachiman protects uninfected bacterial cells in the infected cell population as demonstrated for several novel innate immune systems [[63], [64], [65], [66], [67]]. Since ABI only occurs at a low MOI, this mechanism cannot be inferred to as the major control of phages in environments since the latter outnumber their bacteria prey by 10-fold [[1], [2], [3]]. Since elevated levels of defence system expression facilitates survival of bacterial cells infected with phages at high MOIs, comparative studies of more innate defence systems can yield additional insight into the trade-offs between their fitness cost and antiphage activity.

Data availability statement

All data generated or analyzed during this study are included in this published article and its supplementary information files or are available upon request.

CRediT authorship contribution statement

Xuhui Tian: Writing – original draft, Data curation, Conceptualization. Ruyi Zheng: Data curation. Xin Li: Data curation. Suping Jiang: Data curation. Fang Wang: Writing – review & editing, Supervision. Yulong Shen: Writing – review & editing. Qunxin She: Writing – review & editing, Supervision, Project administration.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Given his role as Executive Editor, Dr. Qunxin She, had no involvement in the peer-review of this article and has no access to information regarding its peer-review. Full responsibility for the editorial process for this article was delegated to Dr. Xiaoxue Wang.

Appendix. Supplementary materials

Supplementary Materials

Supplementary data are available at Engineering Microbiology online.