Phage viruses shape the evolution and virulence of their bacterial hosts. The Salmonella enterica genome encodes several stress-inducible prophages. The Gifsy-1 prophage terminase protein, whose canonical function is to process phage DNA for packaging in the virus head, unexpectedly acts as a tRNase under oxidative stress, cleaving the anticodon loop of tRNA^Leu. The ensuing RNA fragmentation compromises bacterial translation, intracellular survival, and recovery from oxidative stress in the vertebrate host. S. enterica adapts to this tRNA fragmentation by transcribing the RNA repair Rtc system. The counterintuitive translational arrest provided by tRNA cleavage may subvert prophage mobilization and give the host an opportunity for repair as a way of maintaining bacterial genome integrity and ultimately survival in animals.

Assessment of de novo protein synthesis. Protein translation was determined by Western blot version of SUnSET as described previously with minor modifications (33). Briefly, Salmonella stains were grown in MOPS-GLC minimal media pH-7.2 at 37 ºC to an OD₆₀₀ of ~ 0.25 in a shaking incubator. The cells were treated with 400 mM H₂O₂ for 10 min, followed by an additional 40 min incubation in the presence of 250 mg/ml puromycin. Cytoplasmic proteins were extracted and quantified by the BCA method. 30 mg of protein samples were separated on 10% SDS-polyacrylamide gels. The specimens were transferred onto nitrocellulose membranes and immunoblotted using anti-puromycin antibody (Millipore). Protein synthesis was quantified from the immunoblots using the entire molecular weight range of puromycin-incorporated in each lane. The relative puromycin-incorporated levels were quantified with the ImageJ software (NIH). Ponceau stained membrane was used to make sure protein loading was comparable across samples.

Rationale for the H₂O₂ concentrations used in these investigations. The phagosome containing Salmonella is estimated to contain about 16-17 μM H₂O₂ (31). Salmonella containing vacuoles are usually occupied by a single or a handful of bacteria. The in vivo conditions are not conducive to performing biochemical assays. We have determined that, under the highly dense culture conditions used in vitro, higher concentrations of peroxide are needed to elicit responses. The concentrations of 400 μM and 5 mM peroxide were determined empirically in MOPS-GLC media and LB broth, respectively, by testing the bacterial survival and rtcB gene expression (Fig S1D, S3A). 400 μM peroxide was used for the SunSET assays conducted with Salmonella grown in MOPS-GLC media (Fig 1D). We found that the growth of Salmonella was similar in untreated and 400 μM peroxide-treated groups (Fig 1C); however, 400 μM H₂O₂ inhibited de novo translation (Fig 1A, 1B) whereas inducing rtc gene expression (Fig S1D). We tried to recapitulate these conditions to analyze tRNA fragmentation by Northern blots. However, our preliminary studies demonstrated that the quality of total RNA extracted from MOPS-GLC was not satisfactory for Northern blot analysis. Therefore, we grew the bacteria in LB broth, which has shown excellent results in previously published tRNA cleavage work (32). The addition of 5 mM H₂O₂ to 10⁸ Salmonella in LB broth did not affect bacterial viability. Moreover, under these conditions, 5 mM H₂O₂ induced excellent rtcB and rtcR gene transcription. On the other hand, 0.4-1 mM H₂O₂ did not induce rtc gene expression (Fig S3A). Taking into account the preliminary work done to optimize the isolation of high quality tRNA specimens under conditions that do not exert killing of Salmonella while inducing rtc gene transcription, we chose to treat Salmonella grown to high densities in LB broth with 5 mM H₂O₂ (Fig S3A).

Northern blots. Salmonella strains grown in LB broth in a shaking incubator at 37 °C to an OD₆₀₀ of 0.25 were treated with 5 mM H₂O₂ for indicated times. Treated cells were centrifuged at 16,000 × g for 5 min and immediately, total RNA was purified following Trizol method. Total RNA (1 mg) was electrophoresed on 12% acrylamide TBE 8M urea gels and transferred to nylon membrane (Hybond N+, GE) by electroblotting. Membranes were UV-crosslinked (254 nm, 120 mJ dose) twice, blocked in ULTRAhyb-Oligo Buffer (Ambion), and incubated with 5′-³²P-labeled oligonucleotide probes (Table S6) in ULTRAhyb-Oligo buffer at 42°C for 18 h. Membranes were washed with 2X SSC/0.1% SDS washing buffer two times for 10 min each, exposed on a phosphor-imager storage screen, and imaged on a Typhoon 9400 (GE Healthcare).

Animals and animal experiments. C57BL/6J and congenic Cybb^-/- mice deficient in the gp91phox membrane-bound subunit of the phagocyte NADPH oxidase were bred and housed in a specific pathogen-free facility at the University of Colorado, Anschutz Medical Campus. Six to 8-week-old, male and female C57BL/6J and Cybb^-/- mice were inoculated i.p. with ~150 CFU of a Salmonella mixture containing equal numbers of wild-type, and ΔrtcB::km or ΔrtcR::cm Salmonella. The bacterial burden was quantified in livers and spleens 3 days post-infection by plating onto LB agar containing the appropriate antibiotics. Alternatively, in the intestinal infection model, mice were taken off food and water for 4 h prior to p.o. treatment with 20 mg/mouse of streptomycin (Sigma, USA). After 24 h, mice were infected p.o. with equal number of CFUs (2 x 10⁸) of both wild-type and mutant S. enterica. Animals were euthanized 4 d after oral challenge and livers, spleens, and mesenteric lymph nodes tissues were collected, as well as the contents from cecum, colon, and ileum, for quantification of bacterial burden on LB agar plates containing the appropriate antibiotics. Competitive index was calculated as (strain 1 ÷ strain 2)_output ÷ (strain 1 ÷ strain 2)_input. All mice were bred according to protocol #00058 approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Colorado School of Medicine. The macroenvironment is electronically controlled to provide 22.2 ± 1 °C, a 1:10 light/dark cycle, and 30–40% humidity with at least 12 air changes per hour.

Histopathology. Livers and spleens were isolated from C57BL/6J and Cybb^-/- mice 3 days after i.p. inoculation with ~150 CFU of wild-type or DrtcB::km Salmonella. The specimens were fixed in formalin, paraffin-embedded, 4 mm microtome-sectioned, and hematoxylin & eosin-stained. The treatments of the animals and grouping of samples were blinded to the individual scoring the number of microabscesses and necrotic foci in a light microscope. The average number of microabscesses and necrotic foci per 200X field of liver and spleen H&E images were quantified.

H₂O₂ cytotoxicity. Salmonella grown in LB broth in a shaking incubator at 37°C to an OD₆₀₀ of 0.25 was treated with 5 mM H₂O₂ for 30 or 120 minutes. Untreated and treated cells were serially diluted and plated on LB agar for CFU counts. Where indicated, Salmonella grown overnight in LB broth was washed in PBS and resuspended in PBS at 5 x 10⁵ CFU/ml. Cells were treated with 400 mM H₂O₂ for 2 h and plated for CFU counts.

Intracellular survival. Bone marrow-derived macrophages (BMDM) were cultured in L-cell-conditioned medium for 7 days followed by growing in DMEM supplemented with 10% heat-inactivated fetal bovine serum (Thermo Fisher Scientific, Grand Island, NY) and antibiotics. Confluent BMDM cells grown in 96 well plates were infected at an MOI of 2 with Salmonella grown overnight in LB broth at 37 °C in a shaker incubator. Intracellular survival was assessed after cell host lysis by the addition of PBS containing 0.1% Triton X-100. The specimens were serially diluted in PBS, and the number of Salmonella capable of forming a colony in LB agar plates was enumerated after overnight incubation at 37°C. Fold replication was calculated from the number of bacteria recovered after 2 and 18 h of infection compared to time zero.

In vivo ectopic expression of terminase. The gpA gene (STM14_3191) was PCR amplified along with RBS and cloned into pBAD18 and transformed into S. enterica serovar Typhimurium strain 14028s. Bacteria were grown in LB both to an OD₆₀₀ of 0.25 and induced with 0.5% arabinose. Bacteria were simultaneously treated with PBS or 0.5 mM H₂O₂ for 2 h and total RNA was extracted and processed for Northern blotting of tRNA fragments.

Endoribonuclease activity. Salmonella total RNA was extracted from 0.25 OD₆₀₀ cultures as described above. Reaction mixtures were made in 10 ml volumes. 20 pM GpA protein in 20 mM Tris-HCl (pH 7.5) were treated with either 400 μM H₂O₂ or 500 μM DTT and incubated at 37°C for 30 min. Treated proteins were mixed with 2 pM total RNA containing 20 mM Tris-HCl, pH 7.5, 5 mM MnCl₂, 5 mM ZnCl₂ and 0.5 mM ATP and incubated at 37°C for 1 h. The reactions were quenched by adding equal volumes of 90% formamide + 50 mM EDTA, heated at 85°C for 1 minute and electrophoresed through 12% 8M Urea PAGE gels. The RNA was blotted onto Nylon membranes. Northern blots with radiolabelled tRNA probe was performed as before.

In vitro cosN cleavage assay. A 239-bp fragment containing the Gifsy-1 cosN site was amplified from the genome of S. Typhimurium strain 14028s and cloned into HindIII-digested pET22b, creating the pET22-CosN plasmid. 200 ng of cosN-containing or empty plasmids were incubated with recombinant GpA proteins in a reaction buffer containing 50 mM Tris-HCl, pH 8.0, 10 mM MgCl₂, 50 mM NaCl, 5 mM MnCl₂, 5 mM ZnCl₂ and 0.5 mM ATP at 37°C for 60 minutes. The cosN cleavage reactions were terminated by the addition of loading dye containing 25 mM EDTA, and the samples were electrophoresed on 1 % (w/v) agarose gels. Gel images were captured digitally, and cosN cleavage was determined by intensities of linearized and nicked bands over supercoiled form of plasmid.

3’ RACE. S. enterica serovar Typhimurium strain 14028s grown in LB broth in a shaking incubator at 37°C to an OD₆₀₀ of 0.25 was treated with 5 mM H₂O₂ for 2 h. Treated cells were centrifuged at 16,000 × g for 5 min and total RNA was purified following the Trizol method. To determine the site of tRNA fragmentation, total RNA was incubated with GpA protein oxidized with 1 mM H₂O₂ for 2 h. Total RNA was first incubated with 10 mM HCl for 2 h on ice followed by incubation with 5 U of shrimp alkaline phosphatase (NEB) to remove 3′ cyclic phosphates. The RNA was cleaned-up following phenol-chloroform extraction and then ligated to the RNA oligonucleotide 3’-Ada. After ligation, total RNA was treated with DNase I, reverse-transcribed using SuperScript III and oligonucleotide RT, and amplified with oligonucleotides LeuPF and RT (Supplemental Table S6). PCR products cloned using Zero Blunt PCR cloning kit (Invitrogen) were sequenced.

In vitro RtcB ligase assay. Total RNA was treated with H₂O₂-oxidized, recombinant GpA to generate tRNA fragments. The resulting specimens were cleaned up using high pure RNA extraction kit (Sigma, USA). Purified RNA was treated for 1 h with recombinant RtcB (NEB, USA) in the presence of GTP and Mn²⁺ as per supplier’s protocol. The reaction was stopped with 25 mM EDTA. tRNA^LeuPQTV was detected by Northern blot as described above. 

Bioinformatics. Complete protein sequence of Salmonella GpA (STM14_3191) was queried in Jackhmmer for five iterations and 100 representative sequences were retrieved. Similar analysis was done with known tRNA ribonucleases like colicins, Prrc and RNases from Salmonella and other proteobacterial members. The sequences were aligned with COBALT aligner and the result clustal alignment file was uploaded to MegaX software to construct maximum-likelihood phylogenetic tree. Structure of Gifsy-1 GpA was retrieved from the AlphaFold2 database.  The per-residue model confidence score (pLDDT) for Gifsy-1 GpA is 94.65. PDB files were loaded onto ChimeraX program on biopython to view the structures.

Amino acid composition calculation and heatmap construction. Bio.seq module of biopython was utilized to calculate total protein composition and codon adaptation indices for all amino acids from S. enterica serovar Typhimurium 14028s. In brief, parse and seq commands of bio.seqIO were employed to retrieve protein sequences from NCBI entry CP001363. The commands were run on biopython in miniconda environment. The results were extracted as a csv table. Heatmaps were constructed using pheatmap package in R. Proteins with <7.5% leucine content were filtered and gene ontologies were determined by ClueGO analysis (Table S4).

Statistical analysis. Statistical analyses were performed using GraphPad Prism 5.0 software. One-way and two-way ANOVA, t-tests and log-rank tests were used. Data were considered statistically different when p < 0.05.