Various modes of action of dietary phytochemicals, sulforaphane and phenethyl isothiocyanate, on pathogenic bacteria

Isothiocyanates (ITCs) derived from cruciferous plants reveal antibacterial activity, although detailed mechanism is not fully elucidated. Recently it has been reported that ITCs induce the stringent response in Escherichia coli strains. The aim of this work was to determine whether two isothiocyanates, sulforaphane (SFN) and phenethyl isothiocyanate (PEITC), similarly as in E. coli induce stringent response in Bacillus subtilis, model Gram(+) bacterium, and test their potency against a panel of clinical isolates belonging to Gram(+) or Gram(−) groups. Minimal inhibitory concentrations were determined as well as effect of ITCs on membranes integrity, synthesis of DNA, RNA and stringent response alarmones was assessed. SFN and PEITC are effective against B. subtilis and bacterial isolates, namely E. coli, K. pneumonia, S. aureus, S. epidermidis and E. faecalis. Interestingly, in B. subtilis and E. faecalis the inhibition of growth and nucleic acids synthesis is independent of ppGpp accumulation. In bacteria, which do not induce the stringent response in the presence of ITCs, membrane integrity disruption is observed. Thus, ITCs are effective against different pathogenic bacteria and act by at least two mechanisms depending on bacteria species.

S. enterica, Serratia marcescens, Shigella sonnei, Vibrio parahaemolyticus) 10 . Recently, ten ITCs naturally found in cruciferous vegetables, including SFN, AITC, BITC and PEITC, have been tested for 14 bacterial strains growing in BHI medium (B. cereus, B. subtilis, E. faecalis, E faecium, Lactobacillus plantarum, L. monocytogenes, S. aureus, S. xylosus). Results indicated that in most cases bacteria were sensitive to the ITC tested, however the extent of antimicrobial activity depended on the ITC and the strain considered 11 .
Most reports are limited to determination of minimal inhibitory concentration (MIC), which may vary depending on a method used. There is no general rule concerning the efficacy of ITCs toward various bacteria but aromatic ITCs seem to be more active that aliphatic ones 11 . The mechanism of antimicrobial action of ITCs is not well understood. ITC group is highly electrophilic and reacts with amine, thiol or hydroxyl groups, thus it may bind to cellular targets influencing their activity and function. For instance, AITC inhibited activity of thioredoxin reductase and acetate kinase, which are responsible for important metabolic reactions in bacteria 12 , however, existence of ITC-enzyme conjugates has not been proven. It has been also reported that phenyl ITC (PITC), AITC and PEITC had strong antimicrobial activity through the disruption of the bacterial cell membrane function which led to a loss of its integrity and subsequent cell death 13,14 . Recently, it has been shown that PEITC, SFN, AITC, benzyl isothiocyanate (BITC), PITC and isopropyl isothiocyanate (IPRITC) efficiently inhibited growth of Shiga toxin harboring E. coli strains which was accompanied by an inhibition of the lytic development of Shiga toxin-converting bacteriophage and stx gene expression. Moreover, these effects were mediated by alarmones of the stringent response 15,16 .
The stringent response is one of the most far-reaching global regulatory mechanisms in bacteria. The unusual nucleotides, ppGpp and pppGpp (referred together to as (p)ppGpp) are synthesized promptly after the onset of starvation and various physical and chemical stresses. In addition to the stress response, (p)ppGpp regulates also other processes such as biofilm formation, quorum sensing, and bacterial virulence 17,18 . The stringent response is aimed to reprogram cellular metabolism in order to downregulate the active growth and energy consuming processes and to activate those necessary for survival and adaptation to the challenging environmental conditions. For instance, during amino acid starvation, transcription from promoters of genes coding for stable RNA (rRNA or tRNA) or ribosomal proteins is inhibited and transcription from promoters for amino acids transport and biosynthesis is activated 17,19 . One of crucial processes which are affected under conditions of the stringent response is DNA replication. Its downregulation, however, may be achieved at initiation or elongation stage, thus it might not be detectable shortly after stringent response induction but after longer times 20 .
(p)ppGpp metabolism in the cell is under the control of specific enzymes able to synthesize and hydrolyze these nucleotides. E. coli and some of other gamma-and beta-proteobacteria harbor two enzymes, synthetase RelA (specialized in response to amino acid starvation) and bifunctional SpoT, responsible for both synthesis and hydrolysis of (p)ppGpp. While most of other bacterial species, including Gram(+) bacteria, have single bifunctional enzyme, named generally RSH (Rel/Spo homolog) 21 . Nevertheless, in both genetic arrangements, dysfunction of these enzymes is not lethal for bacteria, but leads to the distinct phenotype, named ppGpp-null, characterized by impaired stress response, decreased survival in nutrient limiting conditions and auxotrophy to several amino acids 17,22 .
Despite the presence of (p)ppGpp in majority of free living bacteria, the mechanism of both synthesis and action of this alarmone can differ between bacterial species. In E. coli the (p)ppGpp-mediated regulation takes place mainly at the level of gene expression, with RNA polymerase as a main target of the alarmone direct and indirect effects. On quite a contrary, in many Gram(+) bacteria, including model Bacillus subtilis, (p)ppGpp acts by different strategy; it does not interact directly with RNA polymerase, but its main effect is exerted by controlling the cellular level of GTP 23 . Depleting the cells of GTP, achieved both by lowering the pool of available GTP through (p)ppGpp synthesis and inhibiting the enzymes from GTP synthesis pathway by (p)ppGpp, leads to decreasing of transcription from GTP-starting promoters (including rRNA promoters) 24,25 . This strategy of (p)ppGpp-mediated regulation has been reported for many Gram(+) bacterial species, including E. faecalis, S. aureus, L. monocytogenes and others [26][27][28] .
The purpose of this work was to evaluate the activity of aliphatic (SFN) and aromatic (PEITC) ITC towards different Gram(−) and Gram(+) bacteria and to correlate it with ppGpp production. First, we compared response to ITC of model strains of E. coli and B. subtilis, both wild type and defective in stringent response induction (ppGpp-null variants). Next, we tested activity of ITC toward bacteria belonging to opportunistic human pathogens (K. pneumoniae, S. aureus, S. epidermidis, E. faecalis) ( Table 1). Majority of these species are among the most antibiotic resistant bacteria known, occurring frequently in hospital secondary infections or causing food poisoning.

Results
comparison of response of Escherichia coli and Bacillus subtilis model strains to Sfn or peitc. It has been shown that some ITCs, including SFN and PEITC, induce the stringent response in E.
coli 15,16 . To determine whether tested ITCs have similar mechanism of action in Gram(+) bacteria, we used B. subtilis as a model representative of this group. E. coli cells were more sensitive to tested ITC than B. subtilis: MIC for SFN was 88.6 and 177.3 mg/L, and for PEITC -40.8 and 163.3 mg/L, respectively. Interestingly, E. coli strain unable to produce stringent response alarmones (ppGpp-null) was more sensitive to ITCs than wt strain while sensitivities of wt and ppGpp-null B. subtilis strains were similar (Fig. 1A). It was confirmed by a zone inhibition assay (Fig. 1A,B). Moreover, neither SFN nor PEITC induced (p)ppGpp accumulation in B. subtilis which contrasts with the situation observed in E. coli (Fig. 1C). Interestingly, levels of GTP decreased in both bacteria species when treated with standard stringent control inducer (SHX or RHX) or ITCs, even if (p)ppGpp was not produced in ITC-treated wt strain of B. subtilis (Fig. 1C).
Induction of the stringent response correlates with the inhibition of synthesis of stable RNA and modulation of DNA synthesis. Thus, we compared nucleic acids synthesis in E. coli and B. subtilis and their ppGpp-null counterparts treated or not with ITC at ¼ MIC. In both wild type bacteria strains we observed inhibition of RNA and DNA synthesis upon treatment with SFN, PEITC, and SHX or RHX (positive controls) (Fig. 2). Inhibition of RNA by tested compounds was abrogated in E. coli ppGpp-null cells which proves that it depends on the stringent response. However, synthesis of RNA in B. subtilis was inhibited by ITCs irrespective of the presence of this alarmone ( Fig. 2A). ITC inhibited DNA synthesis to a lesser extent and it was delayed in time as compared to the effects observed for RNA, thus lack of ppGpp had less pronounced effect on this process (at least at tested time  www.nature.com/scientificreports www.nature.com/scientificreports/ points) (Fig. 2B). It is worth to note that SHX or RHX-induced block in nucleic acids synthesis was abrogated in ppGpp-null strains, both E. coli and B. subtilis, which indicates that the stringent response works properly in the wt cells (Fig. 2). The results presented are mean values from 3 independent experiments with error bars indicating SD (error bars are not shown when they are smaller than the symbols). The statistical significance of differences between controls and bacteria treated for 120 min with ITC was determined by student t-test, and asterisks indicate significant differences (***P < 0.001).
www.nature.com/scientificreports www.nature.com/scientificreports/ Lack of the stringent response induction sensitizes E. coli but not B. subtilis to itc. As stringent response protects bacteria against adverse conditions, such as shortage of amino acids, we asked the question about its role in bacteria treated with SFN or PEITC. E. coli and B. subtilis and their ppGpp-null derivatives were treated with ITC at different concentrations (equivalents of ¼, 1, 2, 4x MIC) and their viability was evaluated at different time points. For E. coli lack of ppGpp resulted in an increased sensitivity to high concentrations (2 and 4x MIC) of both tested ITCs (Fig. 3A). Sensitivity of B. subtilis to ITCs did not depend on ppGpp presence which again indicates that response of this bacterium to SFN and PEITC is independent of the stringent response alarmone (Fig. 3B).  Fig. 4. In general, bacterial growth inhibition by ITCs was dose dependent.
To determine whether antibacterial activity of tested ITCs correlates with the stringent response induction in tested bacteria, the levels of (p)ppGpp have been evaluated for selected strains. Results presented in Fig. 5 indicate that SFN and PEITC elevated ppGpp and pppGpp in E. coli, K. pneumonia, S. aureus and S. epidermidis. We did not notice the presence of these alarmones in SFN-or PEITC-treated E. faecalis, although PEITC quite efficiently inhibited growth of this bacterium. SHX, which is a positive control for the stringent response induction, elevated (p)ppGpp in each tested bacterial strain which indicates that stringent response in these isolates functions properly. Interestingly, SFN and PEITC inhibited stable RNA synthesis in all tested bacteria, even in E. faecalis (Fig. 6A). DNA synthesis was also affected although to a different extent by individual compounds which is connected to indirect effect of the inhibition of gene expression (Fig. 6B).
Effect of SFN or PEITC on membrane integrity. It has been previously suggested that ITC may disturb membrane integrity 13 . Our own results indicated that ITCs, including SFN and PEITC, had no significant effect on membrane integrity in E. coli 15,16 . However, our experiments were performed in wt strain of E. coli which was able to induce the stringent response upon ITCs treatment, and the induction of the stringent response may, as an indirect effect, lead to the changes in membrane properties 29 . Thus, we assessed the membrane integrity using propidium iodide test in ppGpp-null E. coli as well as B. subtilis and selected clinically isolated bacteria. As evidenced in Fig. 7A, contrary to wild type E. coli which are impermeable to propidium iodide, the ppGpp-null derivatives stain positively when treated with SFN or PEITC. Interestingly, in B. subtilis the fluorescence signal indicating the disruption of membrane integrity is visible in both, wt and ppGpp-null strains treated with ITCs. It again supports the notion that ITCs do not induce the stringent response in B. subtilis, which would otherwise protect against membrane stress. Similarly, E. faecalis sensitivity to ITCs relies on the loss of membrane integrity probably due to lack of stringent response induction (Fig. 7B). Results presented in Fig. 3S confirm this notion: induction of ppGpp production by RHX protects against membrane permeability induced by ITC in wt B. subtilis.

Discussion
Recently we reported that the antibacterial activity of ITCs against enterohaemorrhagic E. coli relies on the stringent response induction which is connected with massive production of (p)ppGpp, alarmones responsible for, among others, growth retardation and inhibition of stable RNA synthesis or Shiga toxin-converting phage development 15  The statistical significance of differences between controls and bacteria treated for 120 min with ITC was determined by student t-test, and asterisks indicate significant differences: *P < 0.05; **P < 0.01; ***P < 0.001. www.nature.com/scientificreports www.nature.com/scientificreports/ What is interesting, growth inhibition by ITC is independent of (p)ppGpp production in some of tested bacteria. The example is B. subtilis which does not elevate (p)ppGpp levels after ITC treatment. Interestingly, ITCs in both wild type and ppGpp-null strains of B. subtilis inhibit nucleic acids synthesis and cell viability. The explanation for this phenomenon might be a drop in GTP level which we observed in ITCs-treated B. subtilis cells. The regulation of the GTP levels underlies the effects of the stress response of Gram(+) bacteria, e.g. rRNA promoter transcription. The decrease in GTP level under ITC treatment may be responsible for observed RNA synthesis down-regulation. In our case we did not observe (p)ppGpp production in B. subtilis treated with ITCs, thus it seems that GTP drop is mediated by some other mechanisms. Similar effect, i.e. lack of stringent response induction but inhibition of growth and nucleic acids synthesis by ITCs accompanied by decrease in GTP level, we observed in another Gram(+) bacterium, namely E. faecalis.
It has been shown that AITC and PEITC were effective against E. coli, L. monocytogenes, S. aureus and P. aeruginosa, and when used at high concentrations (100, 500 or 1,000 mg/L), they modified physicochemical properties of bacterial surface which led to cytoplasmic membrane permeabilization 14 . Here we confirm that SFN and PEITC compromise the integrity of cytoplasmic membrane, however this is evident only in these strains which do not produce (p)ppGpp upon ITCs treatment, i.e. B. subtilis, E. faecalis and E. coli ppGpp-null variant, and might be the main mechanism underlying antimicrobial activity of ITCs. Interestingly, strains able to induce stringent response, in addition to the observed inhibition of RNA and DNA synthesis, do not react to ITCs treatment with increased membrane permeability, even treated with very high concentrations of ITC (for instance, K. pneumoniae treated with >2 g/L SFN or 1 g/L PEITC). This is in the agreement with reports showing that ppGpp changes the rate of phospholipid synthesis and increases proportion of saturated to unsaturated fatty acids which results in reduction of membrane permeability in amino acid starved E. coli 31,32 and the generalized response to the envelope stress, e.g. in S. aureus 27 , and in E. coli 33 . Such response provides a survival advantage in challenging conditions with the example presented by our results where wild type E. coli exposed to high concentrations of ITCs (2 or 4x MIC) reveals better survival than its ppGpp-null derivative in prolonged tests. The difference in the response to ITCs treatment dependent on the presence of (p)ppGpp indicates that in the wild type cells induction of the stringent response prevents the membrane disruption by ppGpp-mediated gene expression changes leading to the higher tolerance to the envelope stress.
The question arises about the mechanism of species-specific induction of the stringent response by ITCs. Our previous results indicate that supplementation with high concentrations of specific amino acids reversed inhibitory potential of ITCs. For example, supplementation of glycine, arginine, or to lesser extent lysine, methionine, phenylalanine, serine or threonine restored growth of E. coli in the presence of PEITC 15 ; and glycine, cysteine, arginine, tryptophan and aspartic acid decreased sensitivity to SFN 16 . These results suggest that ITCs might bind and titrate these amino acids or block the aminoacylation of relevant tRNA (e.g. by inhibiting a specific aminoacyltransferase), thus elevating their level in cells would prevent the stringent response induction. It is also possible that in some bacteria titration of specific amino acids by ITCs is not a rate limiting step for stringent response induction, however, such hypothesis needs further experimental validation.
Concluding, our results indicate that ITCs may have more than one target in the bacterial cells, and the mechanism of their antibacterial activity is species-specific. Noteworthy, SFN was reported to have various targets in eukaryotic cells 34 and modulate numerous cell signalling pathways, some -specific for particular cell line 2 . Here we show that ITCs act in at least two ways in bacteria, but -interestingly -the outcome is similar, namely bacterial growth inhibition. This is of great importance for the potential use of ITCs as the antimicrobial agents in the pathogenic bacteria infections. The statistical significance of differences between controls and bacteria treated for 50-60 min with ITC was determined by student t-test, and asterisks indicate significant differences: **P < 0.01; ***P < 0.001.

Methods
Bacterial strains and growth conditions. Bacterial strains used in this study are listed in Table 1. All of clinical strains included in this study were isolated from 2009 until 2017 in Voivodship Hospital in Gdańsk. Bacteria were grown in LB, BHI or MH liquid medium in 37 °C with aeration by shaking or plated on solid medium (LB, BHI and MH supplemented with 1.5% bacteriological agar) and incubated overnight in 37 °C. Serine starvation was induced by addition of serine hydroxamate (SHX) or arginine hydroxamate (RHX) to final concentration of 0.5 mg/mL. Susceptibility testing. Antimicrobial activity of SFN and PEITC (LKT Labs, USA) was assessed by determination of the MIC in accordance with CLSI guideline M07-A9. MICs were assessed in two-fold broth microdilution assay and concentrations range from 12.5 to 800 mg/L were used. Zone inhibition tests were assessed by Kirby Bauer disc diffusion method using 6 mm membrane discs (Biomaxima, Poland). Cell concentration of tested microorganisms was adjusted at 0.5 McFarland turbidity standards and inoculated on MHA plates and diameters of growth inhibition were measured after 20-h incubation in 37 °C. www.nature.com/scientificreports www.nature.com/scientificreports/ evaluation of DnA and RnA synthesis. The assessment of nucleic acids synthesis was performed as described 15 . Briefly, bacteria were grown in MH in 37 °C to A 600 of 0.1 and the radioactive precursor of DNA or RNA synthesis, [ 3 H]thymidine and [ 3 H]uridine, respectively, was added at 5 µCi. This time was set as time zero and SFN and PEITC in range of concentrations were added to the cultures. Samples (50 µl) in indicated time were collected on Whatmann 3MM filter papers and transferred to ice-cold 10% trichloroacetic acid (TCA) for 10 min, washed in 5% TCA and in 96% ethanol. After drying of filters, the radioactivity was measured in a scintillation counter MicroBeta (PerkinElmer, Wallac). Results (expressed in cpm) from three independent experiments were normalized to bacterial culture density (A 600 ) and presented as mean values with SD.
Determination of bacterial growth inhibition. Growth inhibition was measured spectrophotometrically (A 600 ) in the presence of relevant concentrations of ITCs. Bacterial cultures were grown in 37 °C to early exponential phase A 600 of 0.1 and treated with indicated compounds. Independent experiments were assessed at least in triplicate and results were expressed as mean ± standard deviation (SD).
Assessment of (p)ppGpp cellular levels. pppGpp and ppGpp levels were measured basically as described 35 with modifications as in 15 . Briefly, overnight bacterial plate culture was resuspended in MOPS low phosphate (0.4 mM) minimal medium with the addition of [ 32 P]orthophosphoric acid (150 μCi/ml) and grown for at least two generations in 37 °C. Then, culture was divided into new flasks and ITCs (1/4xMIC) or SHX, RHX (1.5 mg/ml both) were added at time zero. The untreated culture was used as a negative control. Bacterial samples, collected at time points, were lysed with formic acid (13 M) in three cycles of freeze-thaw procedure. After centrifugation, nucleotides extracts were separated by thin-layer chromatography, using PEI cellulose plates in 1.5 M potassium phosphate buffer. The chromatograms were analysed in phosphorimager (Typhoon, GE Healthcare).
Membrane integrity assay. Bacteria were grown with shaking in 37 °C to early exponential phase A 600 of 0.1 in the same conditions as for growth inhibition experiment. Then, 1 ml of cultures were transferred to Eppendorf tubes 1.5 ml and treated with ITCs in 4xMIC concentration by 1 hour in 37 °C. To determine membrane integrity bacterial cells were washed and resuspended in 1 ml PBS and stained with propidium iodide (IP) 10 µM for 10 min. Bacteria were collected by centrifugation (4,000 × g for 5 min) and washed twice in PBS to remove remaining dye from supernatant. Fluorescence signal from permeabilized cells was analysed using Leica DMI4000B microscope with dedicated filter No. N2.1. Images were collected and processed using LAS AF 3.1 software (Leica). Experiments were repeated at least three times. Results were expressed on photographs as merge of transmitted light-Differential Interference Contrast (DIC) and fluorescence channel.

Data Availability
All data generated or analyzed during this study are included in this published article and Supplementary Information File.