Methylthioadenosine Suppresses Salmonella Virulence

Previously, we demonstrated elevated concentrations of MTA in plasma during systemic infection of mice with S. Typhimurium (17). While the effects of MTA on the host inflammatory response have been documented (16–20), we asked whether elevated extracellular MTA levels could directly impact bacterial virulence. We examined the ability of S. Typhimurium pretreated with MTA (300 μM) to induce pyroptosis and invade human cells. As our original studies that identified MTA to be a modulator of pyroptosis were performed in lymphoblastoid cells (LCLs) and because B cells are a natural target of S. Typhimurium invasion in vivo (21–23), we first looked for an effect of exogenous MTA in 18592 LCLs. We assessed both pyroptosis and invasion by pairing a modified gentamicin protection assay with flow cytometry, as previously described (24) (Fig. 1A). Briefly, pyroptosis was measured by quantifying the number of cells that stained positive for 7-aminoactinomycin D (7-AAD) at 3 h postinfection with S. Typhimurium. Independently, cellular invasion was quantified by using bacteria with a green fluorescent protein (GFP)-harboring plasmid, inducing GFP production after gentamicin treatment, and quantifying the number of GFP-positive (GFP⁺), 7-AAD-negative (7-AAD⁻) host cells at 3 h postinfection. MTA pretreatment had no effect on bacterial growth from the initial overnight culture dilution through late log phase to induce SPI-1 gene expression (Fig. 1B). MTA-pretreated bacteria displayed a 30% decrease (P = 0.0001) in their ability to induce pyroptosis (Fig. 1C). Furthermore, we observed a 35% decrease (P = 0.007) in their ability to invade LCLs (Fig. 1C). We observed similar effects of MTA on pyroptosis and invasion in THP-1 monocytes (Fig. 1D). This was in contrast to our previous finding that MTA treatment of host cells primes them to undergo higher levels of pyroptosis upon S. Typhimurium infection (16) and demonstrates that the molecule has separate effects on both the host and the pathogen.

This reduction in pyroptosis could not be reproduced by treating the bacteria with other metabolites related to the mammalian methionine salvage pathway, including methionine, SAM (also known as AdoMet), α-keto-γ-(methylthio)butyric acid (MTOB), and phenylalanine (Fig. 1E and F). Similarly, adenosine (which lacks only the methylthio group of MTA) was not sufficient to suppress pyroptosis induction. In fact, adenosine increased pyroptosis in both LCLs and THP-1 monocytes. Of note, the bacterial cell is reportedly impervious to SAM (25, 26), so we cannot rule out the possibility that high intracellular concentrations of the molecule could suppress pyroptosis; however, our results rule out the possibility that the molecule is an external signal regulating pyroptosis. Thus, these data demonstrate that MTA exposure uniquely suppresses the ability of S. Typhimurium to induce pyroptosis and invade host cells.

In order to provide independent evidence for MTA-mediated regulation of Salmonella virulence, we genetically disrupted the S. Typhimurium methionine metabolism pathway. The protein MetJ is the master repressor of the methionine regulon and transcriptionally blocks multiple enzymatic steps that enable the generation of methionine and SAM (Fig. 2A) (10). We hypothesized that deletion of metJ would relieve this transcriptional suppression and result in elevated intracellular MTA levels. Therefore, we generated a ΔmetJ mutant and performed mass spectrometry to examine how metabolites in the methionine metabolism pathway were impacted by the mutation. In line with previous reports, we observed an increase in methionine levels in the ΔmetJ mutant (27) (Fig. 2B). MTA, SAM, and phenylalanine levels were also increased (Fig. 2C to E). Expression of metJ from a plasmid reversed these increases (Fig. 2B to E). Consistent with MTA being an inhibitor of polyamine synthesis, the polyamine spermine level was decreased in the ΔmetJ mutant, while no change was observed for another polyamine, spermidine (Fig. 2F and G). Disrupting the methionine metabolism pathway by deleting metJ did not affect bacterial growth (Fig. 2H).

After demonstrating that metJ deletion leads to the accumulation of MTA in the bacterial cell, we examined whether critical Salmonella virulence processes are repressed in the ΔmetJ mutant. Similar to S. Typhimurium treated with MTA, the ΔmetJ mutant had a reduced ability to induce pyroptosis in LCLs and THP-1 monocytes (Fig. 3A and B). Importantly, MTA was the only metabolite whose levels were measured to be elevated in the ΔmetJ mutant that inhibited the levels of pyroptosis when added exogenously to wild-type (WT) bacteria (Fig. 1D). This reduction in pyroptosis was rescued by expressing metJ from a plasmid. S. Typhimurium ΔmetJ had reduced invasion of LCLs, THP-1 monocytes, and HeLa cells (Fig. 3A to C). Spermine and other polyamines could not rescue the pyroptosis induced by S. Typhimurium ΔmetJ, suggesting that these findings are not mediated by the reduction in spermine that we observed in the ΔmetJ mutant (Fig. 3D and E).

Motility and the SPI-1 type III secretion system (T3SS) are critical processes for the induction of pyroptosis and S. Typhimurium invasion in vitro. Motility increases the frequency by which interactions between host and S. Typhimurium cells occur and enables bacterial scanning of the host cell surface to optimize invasion (28, 29). The SPI-1 secretion system not only enables the transport into the host cell of effector proteins that enable invasion (1, 2) but also acts as a trigger for pyroptosis in human cells (30–34). Therefore, our observation that both pyroptosis and invasion are suppressed in MTA-treated S. Typhimurium and S. Typhimurium ΔmetJ led us to hypothesize that motility and/or SPI-1 are suppressed in response to increased concentrations of MTA.

In order to determine whether motility was impaired in the ΔmetJ mutant, we performed a standard bacterial soft agar motility assay (35) and found that the ΔmetJ mutant was able to traverse only two-thirds of the distance of the wild-type bacteria over 6 h (Fig. 4A). This motility defect was restored by expression of metJ from a plasmid. In order to confirm that increased MTA levels were sufficient to drive this phenotype, we also examined S. Typhimurium motility on soft agar containing 300 μM MTA. Like the ΔmetJ mutant, wild-type bacteria swimming on MTA-containing agar demonstrated approximately two-thirds the motility of those on dimethyl sulfoxide (DMSO)-containing agar (Fig. 4B). To further characterize this motility defect, we examined the expression of key motility regulators in the ΔmetJ mutant by quantitative PCR (qPCR). While the master regulator of the flagellar regulon, flhD, did not show reduced expression, two class 2 flagellar genes, fliA and fliZ, showed significant downregulation, which correlates with the decrease in motility (Fig. 4C). Together, these data reveal that MTA suppresses the S. Typhimurium flagellar regulon downstream of flhD transcription, resulting in impaired motility in vitro.

Deploying the SPI-1-encoded T3SS depends on a complex regulatory network in which HilD, HilC, and RtsA drive expression of hilA (36, 37). HilA then enables the expression of the inv-spa and prg-org operons (38–40). Both HilD and HilA directly promote the expression of the first gene in the inv-spa operon, invF (41), which is then able to drive the expression of the sic-sip operon and a number of other critical effectors (39, 42, 43).

Three lines of evidence demonstrated suppression of SPI-1 by MTA. First, we observed suppression of genes carried by SPI-1 in S. Typhimurium ΔmetJ by qPCR. In particular, we saw a decrease in invF expression, as well as a decrease in expression of the translocon component sipB. The finding that sipB is significantly suppressed is of note, as SipB is involved in induction of host cell death by S. Typhimurium (33). Further, while not statistically significant when taking into account multiple-test correction, we saw comparable changes trending toward significance in the expression of the regulatory factor rtsA, the chaperone protein sicP, and the needle complex component prgH (Fig. 5A). This shows that there is at least modest suppression of SPI-1-regulated genes by MTA on the transcriptional level under standard culture conditions. Second, reduced expression of SipA, a SPI-1-secreted effector, was detected in ΔmetJ mutant cell lysates by Western blot staining (Fig. 5B). Third, this reduction in SipA was greater in ΔmetJ mutant cell-free supernatants than cell lysates, suggesting that both expression of SipA and its secretion by the T3SS apparatus are suppressed in the mutant (Fig. 5B). This is consistent with the suppression of invF and sipB observed by qPCR. These reductions in SipA expression were also detected in MTA-treated S. Typhimurium (Fig. 5C). Together, these data demonstrate suppression of the SPI-1 network by MTA.

Our results indicated that both the flagellar regulon and SPI-1 secretion are disrupted in the ΔmetJ mutant. The flagellar regulon helps control SPI-1 gene expression through FliZ-mediated posttranslational HilD regulation (44–46). Likewise, HilD binds the flhDC promoter to activate the flagellar regulon (47, 48). Therefore, our results were compatible with either MTA suppressing the flagellar regulon to suppress the SPI-1 pathway, MTA suppressing the SPI-1 pathway to suppress the flagellar regulon, or MTA suppressing both pathways through independent mechanisms.

If MTA acts through the flagellar regulon, we hypothesized that ablation of the flagellar regulon would prevent the further reduction of pyroptosis and invasion by MTA. However, deletion of metJ resulted in similar reductions in cell death and invasion regardless of the presence or absence of fliZ or flhDC (Fig. 6A and B). Therefore, the flagellar regulon is not necessary for the effects of MTA on pyroptosis and invasion.

If MTA acts through the SPI-1 pathway to suppress the flagellar regulon, we would expect hilD deletion to make the bacteria insensitive to the MTA-mediated suppression of motility. In contrast to this hypothesis, deleting metJ in wild-type and ΔhilD backgrounds resulted in the same decreases in motility (Fig. 6C). Therefore, MTA regulation of SPI-1 is not necessary for the effects of the metabolite on motility.

These findings are consistent with MTA independently regulating the SPI-1 pathway and motility regulon. This is in line with previous observations that while SPI-1 and flagellar gene expression are often correlated, few SPI-1 regulators control invasion by modulating the activity of the flagellar regulon and FliZ (46). We hypothesize that MTA suppresses SPI-1 and motility by interacting with a currently unknown factor upstream of both pathways. Importantly, if MTA interacts with a factor upstream of these virulence pathways, then other virulence pathways may also be suppressed by MTA.

Based on our findings that multiple virulence pathways are disrupted in the ΔmetJ mutant, we hypothesized that the ΔmetJ mutant would have impaired virulence in vivo. To test this, we orally coinfected C57BL/6J mice with wild-type and ΔmetJ mutant S. Typhimurium and measured the bacteria's ability to infect and disseminate to the spleen. We found that the ΔmetJ mutant had a 30-fold reduction in fitness compared to wild-type S. Typhimurium by 5 days postinfection (Fig. 7A).

As the ΔmetJ mutant has reduced SPI-1 secretion, which is critical for Salmonella colonization and dissemination from the mouse gut (2, 8, 36, 49–51), we hypothesized that following oral inoculation, MTA-mediated suppression of SPI-1 was responsible for the reduced ΔmetJ mutant fitness. To test this hypothesis, we knocked out the master SPI-1 regulator, hilD, in both the wild-type and ΔmetJ mutant backgrounds to examine whether SPI-1 was necessary for the reduced ΔmetJ mutant fitness in vivo. Unexpectedly, S. Typhimurium ΔhilD was still able to outcompete ΔhilD ΔmetJ bacteria in the ileum and spleen at 3 days postinfection (Fig. 7B). There were no statistically significant differences between the competitive indexes calculated for the WT versus the ΔmetJ mutant comparisons and the ΔhilD mutant versus the ΔhilD ΔmetJ mutant comparisons in either tissue (P = 0.8 in the spleen and P = 0.9 in the ileum). This demonstrates that MTA suppresses S. Typhimurium virulence independently of its effects on SPI-1.

Our observation that MTA suppresses virulence in an oral model and in an SPI-1-independent manner prompted us to examine whether the ΔmetJ mutant was attenuated through the intraperitoneal (i.p.) route as well. We observed a 5-fold reduction in the fitness of the ΔmetJ mutant relative to that of the wild-type bacteria in the spleen (Fig. 7C). Importantly, the fitness cost was lower than that in spleens from mice orally infected with S. Typhimurium, which suggests that MTA suppresses fitness in an SPI-1-independent manner in the gut as well as by other mechanisms in systemic tissues. Together, these data demonstrate that MTA is able to regulate S. Typhimurium fitness at multiple stages of infection, highlighting the importance of the metabolite's regulation during infection.

We previously reported that treatment of mice with MTA before infection with a lethal dose of S. Typhimurium resulted in reduced production of sepsis-related cytokines (interleukin-6 [IL-6] and tumor necrosis factor alpha [TNF-α]) and modestly prolonged survival (17). In light of our findings here, we hypothesized that the previously observed effects on inflammation may be due to MTA's impact on the microbe.

To test whether MTA reduced host inflammation by suppressing S. Typhimurium proinflammatory genes, we injected mice with a lethal dose (1 × 10⁶ CFU) of either wild-type or ΔmetJ mutant S. Typhimurium by the intraperitoneal route and measured the numbers of CFU and cytokine levels at 4 h postinfection. Similar to what we observed with exogenous treatment of the mice with MTA (17), we did not see a difference in the numbers of CFU, demonstrating that both wild-type and ΔmetJ mutant S. Typhimurium are equally capable of colonizing the spleen at this very early time point (Fig. 8A). In contrast, the IL-6 level was 28% lower (P = 0.03) in mice infected with the ΔmetJ mutant (Fig. 8B). Further, TNF-α concentrations also showed a similar relative decrease (34%), though that result did not meet statistical significance (P = 0.09). Therefore, the ΔmetJ mutant phenocopies the anti-inflammatory effects of treating mice with MTA before infection with S. Typhimurium. Therefore, the anti-inflammatory effects of MTA could be mediated not only through effects on the host but also through suppression of S. Typhimurium proinflammatory virulence factors.

HapMap LCLs were purchased from the Coriell Institute. LCLs and THP-1 monocytes were cultured at 37°C in 5% CO₂ in RPMI 1650 medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS), 2 μM glutamine, 100 U/ml penicillin G, and 100 mg/ml streptomycin. HeLa cells were grown in Dulbecco modified Eagle medium supplemented with 10% FBS, 1 mM glutamine, 100 U/ml penicillin G, and 100 mg/ml streptomycin. Cells used for Salmonella gentamicin protection assays were grown in antibiotic-free medium 1 h prior to infection.

All Salmonella strains are derived from the S. Typhimurium NCTC 12023 (ATCC 14028) strain and are listed in Table 1. All knockout strains were generated by bacteriophage lambda red recombination (75). Recombination events were verified by PCR, and the pCP20 plasmid was used to remove the antibiotic resistance cassette after recombination (76). Strains were cultured overnight in LB broth (Miller), subcultured 1:33, and grown for 2 h 40 min with shaking at 37°C before all experiments, unless otherwise noted. Strains containing the temperature-sensitive plasmid pKD46 or pCP20 were cultured at 30°C and removed at 42°C. Ampicillin was added to LB at 50 μg/ml, kanamycin at 20 μg/ml, and tetracycline at 12 μg/ml. Exogenous metabolites [5′-deoxy-5′-(methylthio)adenosine, l-phenylalanine, l-methionine, α-keto-γ-(methylthio)butyric acid sodium salt, adenosine, spermine, spermidine, and putrescine dihydrochloride (all from Sigma) and S-(5′-adenosyl)-l-methionine chloride (hydrochloride) (Cayman Chemicals)] were added to LB during the 2-h 40-min subculture step at a 300 μM final concentration unless otherwise noted. Infection of cells with metabolite-treated bacteria resulted in a greater than 1:50 dilution of the metabolite.

As previously described, inducible GFP plasmids were transformed into S. Typhimurium strains in order to assess both Salmonella-induced cell death and invasion by flow cytometry (24). Briefly, bacterial cultures were prepared as described above and used to infect LCLs and THP-1 monocytes (multiplicity of infection [MOI], 10 or 30), as well as HeLa cells (MOI, 5). For experiments with nonmotile ΔflhDC mutants, cells were centrifuged at 500 × g for 10 min to enable infection. At 1 h postinfection, cells were treated with gentamicin (50 μg/ml), and IPTG (isopropyl-β-d-thiogalactopyranoside) was added at 2 h postinfection to induce GFP expression. At 3.5 h postinfection, cells were assessed for cell death using 7-aminoactinomycin D (7-AAD; Biomol), with death being read by a Guava EasyCyte Plus flow cytometer (Millipore). Percent invasion was determined by quantifying the number of GFP⁺ 7-AAD⁻ cells at 3.5 h postinfection using the Guava EasyCyte Plus flow cytometer. Gates were set by using uninfected cells to gate out GFP-negative (GFP⁻) cells and by using the natural break in 7-AAD⁻ and 7-AAD-positive (7-AAD⁺) cells (Fig. 1A).

Bacteria were grown overnight as described above, subcultured 1:33 in 10 ml LB, and grown for 2 h 40 min. After thorough washing in phosphate-buffered saline (PBS), samples were flash frozen and thawed, and 0.5 ml PBS was added directly onto the pellets. Samples were then transferred to 2 ml CK01 bacterial lysis tubes (Bertin). These were then taken through 3 cycles of 20-s bursts at 7,500 rpm with 30-s pauses in between bursts using a Bertin Precellys homogenizer (the protocol recommended by Bertin). Samples were spun at 5,000 × g for 5 min, and a Bradford assay was performed on each lysate to gather protein concentration values. One hundred microliters from each homogenate was pipetted directly into a 2-ml 96-well plate (Nunco).

The internal standard methanol solution was made by pipetting 166.7 μl of NSK-A standard (Cambridge Isotope) at 500 μM, 62.5 μl of 500 nM d3-MTA, and 49.771 ml of methanol (MeOH). Nine hundred microliters of this internal standard solution in MeOH was pipetted into all of the standard and sample wells. The plate was then capped and mixed at 700 rpm at 25°C for 30 min. The plate was then centrifuged at 3,000 rpm for 10 min. Using an Integra Viaflo96 pipette, 600 μl of extract was pipetted out and transferred to a new 96-well plate. The extracts were allowed to dry under a gentle stream of nitrogen until completely dry. Thirty-two microliters of 49:50:1 water-acetonitrile-trifluoroacetic acid was added to each well and mixed at 650 rpm for 10 min at room temperature. Then, 128 μl of 1% trifluoroacetic acid was added to each well, the contents were mixed briefly, and the plate was centrifuged down to give a total of 160 μl of sample.

The samples were analyzed using ultraperformance liquid chromatography-electrospray ionization-tandem mass spectrometry (UPLC-ESI-MS/MS) using a customized method allowing chromatographic resolution of all analytes in the panel. Flow from the liquid chromatography separation was introduced via positive-mode electrospray ionization (ESI⁺) into a Xevo TQ-S mass spectrometer (Waters) operating in multiple-reaction-monitoring (MRM) mode. MRM transitions (compound-specific precursor-to-product ion transitions) for each analyte and internal standard were collected over the appropriate retention time. The data were imported into Skyline software (https://skyline.gs.washington.edu/) for peak integration and exported into Excel software for further calculations.

Bacteria were grown as described above, and RNA was isolated from 5 × 10⁸ bacteria using the Qiagen RNAprotect Bacteria reagent and an RNeasy minikit (Qiagen) according to the manufacturer's instructions. RNA was treated with DNase I (NEB), and 500 ng was reverse transcribed using an iScript cDNA synthesis kit (Bio-Rad Laboratories). qPCR was performed using the iTaq Universal SYBR green Supermix (Bio-Rad Laboratories). Ten-microliter reaction mixtures contained 5 μl of the supermix, a final concentration of 500 nM each primer, and 2 μl of cDNA. Reactions were run on a StepOnePlus real-time PCR system (Applied Biosystems). The cycling conditions were as follows: 95°C for 30 s, 40 cycles of 95°C for 15 s and 60°C for 60 s, and 60°C for 60 s. A melt curve was performed in order to verify single PCR products. The comparative threshold cycle (C_T) method was used to quantify transcripts, with the ribosomal rrs gene serving as the endogenous control. ΔC_T values were calculated by subtracting the C_T value of the control gene from the C_T value of the target gene, and the ΔΔC_T value was calculated by subtracting the wild-type ΔC_T value from the mutant ΔC_T value. Fold change represents 2^−ΔΔCT. Experiments included three technical replicates, and the data represent the qPCR results from five separate RNA isolation experiments. The oligonucleotides used are listed in Table 2.

Bacteria were grown as described above. For analysis of cell lysates, bacterial cultures were centrifuged at 10,000 × g for 5 min. The supernatant was discarded, and the pellets were lysed in 2× Laemmli buffer (Bio-Rad) with 5% 2-mercaptoethanol. Samples were boiled for 10 min and analyzed on Mini-Protean TGX stain-free gels (Bio-Rad). Bands were stained with a rabbit anti-SipA antibody overnight at 4°C. Antibodies were then detected by staining with the Li-Cor IRDye 800CW donkey anti-rabbit immunoglobulin antibody. SipA was quantified using a Li-Cor Odyssey Fc imaging system paired with Image Studio software. Bands were normalized to total protein using a TGX stain-free system. Total protein was quantified with Fiji software (77).

For secreted protein analysis, cultures were centrifuged at 10,000 × g for 5 min, and supernatants were passed through a 0.2-μm-pore-size syringe filter. At this point, 6 μl of 100-ng/μl bovine serum albumin (BSA) was added to 600 μl of supernatant as a loading control. Chilled 100% trichloroacetic acid (TCA) was added to a final concentration of 10%, and the mixture was incubated on ice for 10 min. Six hundred microliters of chilled 10% trichloroacetic acid was added, and the solution was incubated on ice for another 20 min before being centrifuged at 20,000 × g for 30 min. Pellets were washed twice with acetone and resuspended in 2× Laemmli buffer (Bio-Rad) with 5% 2-mercaptoethanol before boiling for 10 min. Proteins were then analyzed as described above.

The motility assay was performed as previously described (35). Briefly, strains were cultured overnight in LB broth (Miller), subcultured 1:33, and grown for 2 h 40 min with shaking at 37°C. Two microliters of the subcultured solution was plated in the center of a 0.3% agar LB plate supplemented with 50 μg/ml ampicillin. Metabolites or DMSO was added to the solution prior to the agar solidifying in order to allow exposure of the bacteria to the metabolite for the entirety of the assay. The plates were incubated at 37°C for 6 h before the halo diameter was quantified.

Mouse studies were approved by the Duke Institutional Animal Care and Use Committee and adhere to the guidelines in the Guide for the Care and Use of Laboratory Animals of the National Research Council (78). Bacteria were grown as described above, washed, and resuspended in PBS. Inocula were confirmed by plating for determination of the number of CFU. For oral infections, age- and sex-matched 7- to 16-week-old C57BL/6J mice were fasted for 12 h before infection. At 30 min prior to infection, mice received 100 μl of a 10% sodium bicarbonate solution by oral gavage. Age- and sex-matched mice were then infected with 10⁶ bacteria in 100 μl PBS by oral gavage. At 5 days postinfection, mice were euthanized by CO₂ asphyxiation, and spleens and ileums were harvested, homogenized, weighed, and plated on LB agar containing either ampicillin or kanamycin. For ΔhilD mutant experiments, mice were infected with 10⁹ bacteria as described above and harvested 3 days postinfection.

For intraperitoneal (i.p.) competitive index experiments, age- and sex-matched 6- to 20-week-old C57BL/6J mice received 10³ bacteria in 100 μl of PBS by i.p. injection. At 3 to 5 days postinfection, mice were euthanized by CO₂ asphyxiation, blood was drawn by cardiac puncture, and spleens were harvested, homogenized, weighed, and plated on LB agar containing ampicillin or kanamycin.

The competitive index was calculated as (number of ΔmetJ mutant CFU/number of WT CFU)/(number of ΔmetJ mutant CFU in the inoculum/number of WT CFU in the inoculum). Statistics were calculated by log transforming this ratio from each mouse and comparing the value to an expected value of 0 using a one-sample t test. Differences between competitive indexes were calculated by performing a Student's t test comparing the log-transformed competitive indexes.

For high-dose i.p. injection, age- and sex-matched 6- to 8-week-old C57BL/6J mice received 10⁶ bacteria in 100 μl of PBS by i.p. injection. At 4 h postinfection, mice were euthanized by CO₂ asphyxiation, blood was drawn by cardiac puncture, and spleens were harvested, homogenized, weighed, and plated on LB agar containing ampicillin. Plasma was isolated using plasma separation tubes with lithium heparin (BD). IL-6 and TNF-α were then quantified from plasma and spleen extracts using DuoSet enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems).