The Disulfide Stress Response and Protein S-thioallylation Caused by Allicin and Diallyl Polysulfanes in Bacillus subtilis as Revealed by Transcriptomics and Proteomics

Garlic plants (Allium sativum L.) produce antimicrobial compounds, such as diallyl thiosulfinate (allicin) and diallyl polysulfanes. Here, we investigated the transcriptome and protein S-thioallylomes under allicin and diallyl tetrasulfane (DAS4) exposure in the Gram-positive bacterium Bacillus subtilis. Allicin and DAS4 caused a similar thiol-specific oxidative stress response, protein and DNA damage as revealed by the induction of the OhrR, PerR, Spx, YodB, CatR, HypR, AdhR, HxlR, LexA, CymR, CtsR, and HrcA regulons in the transcriptome. At the proteome level, we identified, in total, 108 S-thioallylated proteins under allicin and/or DAS4 stress. The S-thioallylome includes enzymes involved in the biosynthesis of surfactin (SrfAA, SrfAB), amino acids (SerA, MetE, YxjG, YitJ, CysJ, GlnA, YwaA), nucleotides (PurB, PurC, PyrAB, GuaB), translation factors (EF-Tu, EF-Ts, EF-G), antioxidant enzymes (AhpC, MsrB), as well as redox-sensitive MarR/OhrR and DUF24-family regulators (OhrR, HypR, YodB, CatR). Growth phenotype analysis revealed that the low molecular weight thiol bacillithiol, as well as the OhrR, Spx, and HypR regulons, confer protection against allicin and DAS4 stress. Altogether, we show here that allicin and DAS4 cause a strong oxidative, disulfide and sulfur stress response in the transcriptome and widespread S-thioallylation of redox-sensitive proteins in B. subtilis. The results further reveal that allicin and polysulfanes have similar modes of actions and thiol-reactivities and modify a similar set of redox-sensitive proteins by S-thioallylation.

its Zn-binding active site, leading to its inactivation and methionine auxotrophy in HOCl-treated cells [27,35]. The redox-sensing OhrR repressor is inhibited by S-bacillithiolation under HOCl and cumene hydroperoxide (CHP) stress, resulting in derepression of the ohrA peroxiredoxin, which confers resistance to the oxidants [27,29,36]. However, it is unknown whether garlic sulfur compounds modify similar targets by S-thioallylation in B. subtilis.
Here, we aimed to investigate the regulatory stress responses and targets for S-thioallylations in response to allicin and diallyl tetrasulfane (DAS4) in B. subtilis. Both sulfur compounds allicin and DAS4 were shown to elicit a similar strong thiol-specific oxidative and sulfur stress response in the transcriptome of B. subtilis. About 108 targets for S-thioallylation were identified by shotgun proteomics, including the majority of previously identified S-thiolated proteins under HOCl stress, such as TufA, MetE, YxjG, GuaB, SerA, and PpaC, as well as the redox-sensing regulators OhrR, HypR, YodB, and CatR. Growth comparisons revealed that BSH and the OhrR, PerR, HypR, and Spx regulons contribute to allicin protection mechanisms in B. subtilis.
For the stress experiments with allicin and DAS4, cells were grown in BMM to an optical density at 500 nm (OD 500 ) of 0.4 and exposed to sub-lethal doses of 90 and 250 µM allicin and 92 µM DAS4. The statistics of significant changes in the growth curves was determined using the Student's unpaired two-tailed t-test by the GraphPad Prism software. Allicin was synthetized by oxidation of 3-[prop-2-en-1-yl) disulfanyl] prop-1-ene (diallyl disulfide) with peracetic acid, as described [24]. DAS4 was synthesized and purified, as previously described [9].

Identification of S-Thioallylated Proteins Using LTQ-Orbitrap Mass Spectrometry
B. subtilis 168 was grown in BMM and treated with 90 µM allicin and 92 µM DAS4 for 30 min, followed by harvesting of cells, and alkylation in N-ethylmaleimide (NEM) buffer, as described [26,27]. NEM-alkylated protein extracts were subjected to tryptic in-gel-digestion and LTQ Orbitrap Velos mass spectrometry, as described [27]. S-thioallylated proteins were identified by searching all tandem mass spectrometry (MS/MS) spectra against the B. subtilis 168 target-decoy protein sequence database extracted from UniprotKB release 12.7 (UniProt Consortium, Nucleic acids research 2007, 35, D193-197) using Sorcerer TM -SEQUEST ® (Sequest v. 2.7 rev.11, Thermo Electron, including Scaffold 4.0; Proteome Software, Inc., Portland, OR, USA). The SEQUEST search was carried out with the previously used parameters [27], including a parent ion mass tolerance of 10 ppm and a fragment ion mass tolerance of 1.00 Da. Up to two tryptic mis-cleavages were allowed. Methionine oxidation (Met+15.994915 Da), cysteine alkylation by N-ethylmaleimide (Cys+125.04767 Da), and cysteine S-thioallylation by allicin (Cys+72.00337 Da for C 3 H 5 S 1 ) were set as variable modifications. The mass spectrometry data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository [40,41] with the dataset identifier PXD013607.

Microarray Transcriptome Analysis
For microarray analysis, B. subtilis wild-type cells were grown in minimal medium to OD 500 of 0.4 and harvested before and 30 min after treatment with 90 µM allicin and 92 µM DAS4. Total RNA was isolated by the acid phenol method as described [42]. For transcriptome analysis, 35 µg RNA were DNase-treated using the RNase-Free DNase Set (Qiagen, Hilden, Germany) and purified using the RNA Clean-Up and Concentration Kit (Norgen Biotek, Thorold, ON, Canada). The quality of the RNA preparations was assessed by means of the Agilent 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). Fluorescently labeled cDNA was synthesized and purified as described previously [27,43]. The allicin and DAS4 samples were labeled with Cy5, and the control samples were labeled with Cy3. 600 ng of Cy5-and Cy3-labeled cDNA were co-hybridized in a 1:1 ratio with the microarray based on the instruction of Agilent's protocol (Two-Color Microarray-based Gene Expression Analysis, version 5.5, Agilent Technologies, Waldbronn, Germany). Data were extracted and processed using the feature extraction software (version 10.5, Agilent Technologies, Waldbronn, Germany). The error-weighted average of the log ratios of the probes was calculated for each gene using the Rosetta Resolver software (version 7.2.1, Rosetta Biosoftware, Seattle, WA, USA). Normalization was applied to log ratios by using to the Lowess algorithm. Genes showing induction or repression ratios of at least three-fold in two independent biological replicates were considered as significantly induced and subsets of the most interesting regulons are displayed in the Voronoi transcriptome treemap. All transcriptional fold-changes and log2 fold changes of the protein-coding genes and other RNA features quantified for DAS4 or allicin stress versus the control samples including the standard deviations and coefficient of variations are listed in Tables S1 and S2. The microarray datasets are available in NCBI's gene expression omnibus (GEO) database under accession number [GSE132981].

Construction of the Voronoi Transcriptome Treemap
For construction of the allicin and DAS4 transcriptome treemaps, the Paver software (DECODON GmbH, Greifswald, Germany) was applied [44]. The treemap visualizes the log2 fold-changes of highly upregulated redox regulons under allicin and DAS4 stress using a red-blue color gradient. Regulons are indicated with larger white labels, genes and operons are shown with smaller labels. The cell size is defined as ratio of expression levels under allicin treatment relative to the control.

Immunoprecipitation (IP) and Non-Reducing SDS-PAGE Analysis of OhrR-FLAG, HypR, YodB, and CatR Proteins
The OhrR-FLAG protein expressing B. subtilis strain HB9121 was grown in BMM and exposed to 90 µM allicin at an OD 500 of 0.4. Cells were harvested before (as untreated control) and 30 min after allicin stress in TE-buffer (10 mM Tris-HCl, pH8; 1 mM EDTA) with 100 mM iodoacetamide. Alkylated protein extracts were used for IP of OhrR-FLAG protein using anti-FLAG M2-affinity agarose (Invitrogen) according to the instructions of the manufacturer. For IP of HypR, YodB, and CatR, protein extracts of allicin-treated cells were subjected to Dynabead Protein A sepharose coupled to polyclonal HypR, YodB, and CatR antibodies, as described previously [27,45]. The precipitated OhrR-FLAG, HypR, YodB, and CatR proteins were eluted by boiling in non-reducing SDS sample buffer (4% SDS; 62.5 mM Tris-HCl pH 8.0, glycerol) and separated using 15% non-reducing SDS-PAGE. The protein bands were cut from the SDS-gel, tryptic in-gel digested, and the peptides analyzed by Orbitrap mass spectrometry as described above.

Determination of Sub-Lethal Allicin and DAS4 Concentrations and Allicin Priming Assays in B. subtilis
First, we analyzed the growth of B. subtilis wild type cells after treatment with allicin and diallyl tetrasulfide (DAS4) to determine sub-lethal concentrations. Exposure of exponentially growing B. subtilis cells to 90 µM and 250 µM allicin resulted in a dose-dependent lag of growth for 20 min and 2 h, respectively, followed by rapid resumption of growth with the same rate as the untreated control ( Figure 1A). This indicates that B. subtilis cells are able to recover fast in growth, presumably due to rapid detoxification of allicin and DAS4. We were further interested whether low doses of allicin can prime B. subtilis cells to mediate protection against subsequent higher doses of allicin or other oxidants. Indeed, priming of B. subtilis cells with 90 µM allicin resulted in protection against subsequent treatment with lethal 250 µM allicin, as shown by the faster growth recovery in primed cells compared to those treated with 250 µM allicin alone ( Figure 1A). In addition, allicin primed B. subtilis cells could recover from lethal oxidative stress provoked by 250 µM CHP ( Figure 1B). However, allicin priming did not confer cross-protection to lethal doses of 10 mM H 2 O 2 ( Figure 1C). Thus, allicin priming mediates protection against higher allicin doses and strong oxidants, such as CHP in B. subtilis. Next, we analyzed the growth of B. subtilis after exposure to different doses of DAS4. Treatment of B. subtilis with 23, 46, and 92 µM DAS4 also caused a dose-dependent growth delay of 1-2 h, followed by fast recovery of growth ( Figure 1D), similarly as measured after allicin stress ( Figure 1A). These growth profiles after allicin and DAS4 stress seem to indicate a disulfide stress response and are very similar, as shown previously for diamide and NaOCl stress in B. subtilis and for allicin in E. coli [7,27,46]. subtilis 168 was grown in Belitsky minimal medium (BMM) to an OD 500 of 0.4 and exposed to 90 µM allicin for 30 min before subsequent treatment with the higher dose of 250 µM allicin. The growth was improved in the allicin-primed cells compared to non-primed cells, which were treated only with 250 µM allicin. (B) For allicin-CHP cross priming experiments, allicin-primed cells were exposed subsequently to lethal 250 µM CHP stress. Growth was improved in allicin-primed cells compared to non-primed cells. (C) For allicin-H 2 O 2 cross priming, allicin-primed cells were treated with lethal 10 mM H 2 O 2 , but allicin priming did not improve growth. (D) B. subtilis 168 was exposed to different doses of 23, 46, and 92 µM DAS4 leading to different lag phases in growth and recovery after different times. The results are from three biological replicates. Error bars represent the standard deviations (SD) and the statistics was calculated using the Student's unpaired two-tailed t-test by the GraphPad Prism software.

Allicin and DAS4 Cause a Strong Thiol-Specific Oxidative, Disulfide, and Sulfur Stress Response in the Transcriptome of B. subtilis
To investigate in more detail the allicin-and DAS4-induced disulfide stress responses, B. subtilis was exposed to sub-lethal doses of 90 µM allicin or 92 µM DAS4 for 30 min and the changes in the transcriptome were analyzed using DNA microarrays, as described [27]. In total, 515 and 616 genes were reproducibly >3-fold up-regulated in the transcriptomes under allicin and DAS4 stress, respectively, in 2 biological replicates (Tables S1 and S2). The genes were sorted into regulons and subsets of the most strongly induced regulons under allicin and DAS4 stress are displayed in Voronoi transcriptome treemaps (Figure 2A,B, Tables S1 and S2). In general, both allicin and DAS4 caused a very similar stress response in the transcriptomes of B. subtilis. The OhrR, Spx, PerR, HypR, YodB, CatR, AdhR, ArsR, CzrA, CsoR, and CtsR regulons were most strongly induced under allicin and DAS4 stress, indicating that allicin and polysulfanes elicit a strong thiol-specific oxidative, disulfide and metal stress response, as well as protein damage (Figure 2A,B). Thus, these allicin and DAS4 expression profiles are similar compared to the NaOCl transcriptome signature in B. subtilis [27]. both provoke a thiol-specific oxidative stress signature, as revealed by up-regulation of the OhrR, PerR, Spx, HypR, YodB, CatR, CtsR, and CymR regulons. The metal sensing CsoR, ArsR, and CzrA regulons were further induced. All fold-changes of gene expressions were quantified using microarrays and the data is listed in Tables S1 and S2.
The OhrR-controlled ohrA peroxiredoxin gene was among the most highly induced genes under allicin and DAS4 stress (log2 fold-changes 6.4-8.7) (Tables S1 and S2). The OhrR repressor is redox-controlled by S-bacillithiolation under CHP and HOCl stress, leading to derepression of ohrA, which confers resistance to organic hydroperoxides and HOCl [27,29,47,48]. These results suggest that OhrR could sense allicin and DAS4 via S-thioallylation of its lone redox-sensing Cys22 residue. Furthermore, the PerR regulon was up-regulated under allicin and DAS4 stress, including the genes for catalase katA (5.8-7.8 log2 fold-changes), peroxiredoxins ahpCF (3.8-4.8-fold), and the miniferritin mrgA (5.5-8.6), which are indicative of an oxidative stress response. The PerR regulon is also induced by other disulfide stress conditions, such as NaOCl and diamide [27,46]. The majority of genes controlled by the disulfide stress specific Spx transcription factor displayed elevated expression under allicin and DAS4 exposure. These include several genes for thiol-disulfide oxidoreductases (nfrA, yugJ, ywcH and yjbH), thioredoxin/ thioredoxin reductase (trxA, trxB), methionine sulfoxide reductases (msrA, msrB) and other redox enzymes that are required to maintain cellular redox homeostasis [49,50] and displayed log2 fold-changes of 4.2-5.8 under allicin and DAS4 (Figure 2A,B, Tables S1 and S2). These redox enzymes could be involved in detoxification of allicin and polysulfanes or reduction of S-thioallylations to restore the BSH redox balance and reduced protein thiols. In addition, we noted the log2-fold changes of 1.5-3 for the Spx regulon genes bshB2, bshC, brxA, brxB, and ypdA, which encode the pathways for BSH biosynthesis, reduction of bacillithiol disulfide, and regeneration of S-thiolated proteins [19,28,51,52]. Similarly, the CymR regulon for cysteine biosynthesis was weakly induced (log2 fold changes of 1-2) by allicin and DAS4 (Tables S1 and S2). The induction of the genes for BSH and Cys biosynthesis supports the depletion of these LMW thiols in B. subtilis.
Allicin and DAS4 are reactive sulfur species (RSS), and their degradation leads to formation of other RSS, such as allyl thiols, allyl persulfides, and H 2 S [3,5,62]. Thus, allicin and DAS4 might also cause induction of RSS-specific regulons, such as sulfur transferases. Interestingly, we noted the very strongly induced yrkEFHIJ operon, exhibiting log2-fold changes of 6.5-11 under allicin and DAS4 (Figure 2A,B, Tables S1 and S2). The yrkE gene encodes a putative sulfur transferase, the yrkFJ genes share homology to genes encoding sulfur carrier protein subunits of the TusA family, and yrkJ could encode a sulfonate uptake permease. Thus, this yrkEFHIJ operon could function as novel sulfur-specific uptake and degradation operon. This sulfur-specific operon is connected to the YrkP regulon consisting of the yrkON, yrkPQR, and ykcBC operons, which are also weakly up-regulated by garlic sulfur compounds (log2 fold change of 1-2).
Apart from these thiol-stress responses, allicin and DAS4 caused the induction of the SOS response LexA regulon, revealing a DNA damage response in B. subtilis. Increased transcription of the SigD and SigM regulons in response to allicin and DAS4 stress was further noted. The SigD regulon controls motility and chemotaxis, and was previously induced under disulfide stress [27]. The SigM regulon was shown to respond to cell wall antibiotics, ethanol, heat, acid, and superoxide stress [63]. In addition, we noticed that the carbon catabolite control CcpA regulon responds moderately to allicin and more strongly to DAS4 stress, which could point to the utilization of allicin or its degradation products allyl thiols as alternative carbon sources (Figure 2A,B, Tables S1 and S2). Of note was particularly the rbsRKDACB operon, that encodes the ribose uptake ABC transporter with log2 fold-changes of >6 for rbsA, rbsB, and rbsC. In conclusion, allicin and DAS4 lead to a very similar thiol-specific oxidative, disulfide and sulfur stress response and protein damage in B. subtilis, which is comparable to the disulfide stress responses caused by other thiol-reactive compounds, such as diamide and HOCl. These transcriptomics signatures for allicin and DAS4 are in agreement with their main mode of actions to impair the thiol-redox homeostasis by S-thioallylation of LMW and protein thiols [7,9,19].

Allicin and DAS4 Lead to Widespread S-Thioallylation of Total 108 Proteins in B. subtilis
Allicin has been previously shown to modify numerous cytoplasmic proteins by S-thioallylations in E. coli, S. aureus and human Jurkat cells [7,19,26]. Here, we were interested to identify the targets of S-thioallylation in the proteome of B. subtilis. We used Orbitrap LC-MS/MS analysis to investigate the S-thioallylome in B. subtilis after exposure to 90 µM allicin and 92 µM DAS4 stress. S-thioallylated proteins were identified by a mass increase of 72 Da at Cys peptides (Tables S3-S6). Both allicin and DAS4 resulted in a large extent of 108 S-thiolated proteins in the proteome. While 89 Cys residues in 79 proteins were S-thioallylated by allicin, DAS4 treatment resulted in S-thioallylation of 76 Cys residues in 66 proteins. The majority of 53 Cys residues in 44 proteins were modified by both allicin and DAS4. The S-thioallylated proteins were allocated to functional categories based on TIGRfam annotation in B. subtilis, such as information processing (e.g., transcription, protein synthesis), biosynthesis of amino acids, cofactors and nucleotides, energy metabolism, and adaptation to environmental changes (Table S4). The targets for S-thioallylations by allicin or DAS4 were color-coded in the Voronoi proteome treemap based on the abundance of detected S-thioallylated peptides using spectral counts (Figures 3 and 4).
In total, we could quantify about 1137 proteins using spectral counts with the Proteome software Scaffold in the proteome of allicin and DAS4-treated cells of B. subtilis (Table S5). The most abundantly S-thioallylated proteins under allicin and DAS4 stress were protein translation factors (TufA, FusA, Tsf), the surfactin synthetase subunits (SurfAA, SurfAB) and enzymes for amino acid biosynthesis, such as methionine (MetE, YxjG, YitJ, MetI, MtnA, MtnK), arginine (AspB, ArgG, ArgJ, ArgF, CarB), glutamine (GlnA), serine (SerA), and aromatic amino acids (AroA) (Tables S3-S6). Among these, TufA, FusA, and MetE were modified at their conserved Cys residues, e.g., Cys83 of TufA, Cys239 of FusA, and Cys719/Cys730 of MetE. This indicates that reactive garlic compounds target most strongly abundant thiol-containing proteins with conserved Cys-residues. Interestingly, TufA, FusA, and GlnA were also S-thioallylated at the same conserved Cys residues in the allicin proteome of S. aureus [19]. In addition, proteins involved in nucleotide biosynthesis, such as purine and pyrimidine biosynthetic enzymes (PurB, PurC, PurF, PyrAB), the IMP dehydrogenase GuaB, and the manganese-dependent inorganic pyrophosphatase PpaC were S-thioallylated by allicin and DAS4 in B. subtilis. GuaB and PpaC were previously identified as allicin targets in the proteome of S. aureus [19]. In B. subtilis, GuaB and PpaC were S-thioallylated at their conserved redox-sensing active sites Cys308 and Cys158, respectively. Overall, TufA, MetE, YxjG, GuaB, SerA, and PpaC were modified at their redox-sensitive Cys residues by allicin and DAS4, which were previously identified as targets for S-bacillithiolation under HOCl stress in B. subtilis (Tables S3-S6) [27,30,31]. The allicin and DAS4 targets included the peroxiredoxin AhpC and the methionine sulfoxide reductase MsrB, which were S-thioallylated at conserved Cys residues. In conclusion, the reactive sulfur compounds of garlic allicin and polysulfides modify largely conserved redox-sensitive active site Cys residues in the proteomes of S. aureus and B. subtilis.
However, we did not find S-thioallylated redox-sensitive regulators in our proteome dataset using the shotgun proteomics approach, which might be related to their low abundance in the proteome. Based on our transcriptome results, allicin and DAS4 resulted in strong induction of the OhrR, HypR, YodB, and CatR regulons in B. subtilis, which are controlled by redox-sensing MarR/OhrR-type and MarR/DUF24 family regulators. Thus, we were interested whether these MarR-type repressors sense garlic compounds by S-thioallylation. We used immunoprecipitation to pull-down OhrR-FLAG protein with anti-FLAG agarose from cell extracts of allicin-treated B. subtilis strain HB9121. Allicin-treated cell extracts of B. subtilis 168 were used to pull-down HypR, YodB, and CatR proteins with polyclonal antibodies. Orbitrap mass spectrometry of the IP samples enabled the identification of conserved redox-sensing Cys residues as modified by S-thioallylation for OhrR (Cys15), HypR (Cys14), and CatR (Cys7) (Tables S3-S6). For the YodB-repressor, only Cys101 was S-thioallylated, but the N-terminal redox-sensitive Cys6 peptide could not be identified. Thus, our results confirm that the redox-sensitive MarR-type repressors OhrR, CatR, and HypR sense allicin by S-thioallylation, resulting in repressor inactivation and induction of the peroxidiredoxins (OhrA) and flavin disulfide reductases (HypO) that might function in allicin detoxification in B. subtilis. Similarly, the MarR/SarA family regulators MgrA, SarA, and SarS were previously shown to be S-thioallylated by allicin and MgrA was S-sulfhydrated at the redox-sensing Cys12 in response to H 2 S in the proteome of S. aureus [19,64]. Overall, we have identified 108 S-thioallylated proteins in the proteome of B. subtilis in response to allicin and DAS4 stress. These contain 42 Cys peptides with conserved Cys residues as revealed by the conserved domain database (Table S4). Thus, the main targets for S-thioallylations in B. subtilis are translation elongation factors (TufA, Tsf, FusA), redox-sensing MarR-type regulators (OhrR, HypR, CatR), many biosynthetic enzymes (GuaB, MetE, YxjG, SerA, PpaC), and antioxidant enzymes (AhpC, MsrB) that harbor conserved active site Cys residues and overlap strongly with the allicin targets in the proteome of S. aureus [19].

The LMW Thiol Bacillithiol and the Redox-Sensitive Regulators OhrR, HypR, and Spx Functions in the Defense of B. subtilis Against Allicin Stress
Next, we were interested whether the LMW thiol BSH and the antioxidant enzymes controlled by redox-sensitive regulators OhrR, HypR, and Spx provide protection against allicin stress in B. subtilis. Thus, growth curves of the bshA, ohrA, hypR, and spx mutants were monitored compared to the wild type after exposure to sub-lethal 90 µM allicin stress. Allicin resulted in a short lag of growth for 60 min in the wild type, followed by fast resumption of growth with the same growth rate as untreated control cells. In contrast, the bshA mutant was impaired in growth after exposure to 90 µM allicin ( Figure 5A). Treatment of the ohrA and hypR mutants with 90 µM allicin also led to a reduced growth rate and delayed resumption of growth after 180 min ( Figure 5B,C). Thus, the LMW thiol BSH, the OhrR and HypR regulons provide protection against allicin toxicity in B. subtilis, possibly by detoxification of allicin. The spx mutant displayed a reduced growth rate even under non-stress conditions and was unable to grow with 90 µM allicin compared to the wild type ( Figure 5D). These results indicate a role of the Spx regulon to combat thiol-stress provoked by allicin in B. subtilis. Figure 5. The bshA, ohrA, hypR, and spx mutants show growth defects under allicin stress in B. subtilis. Growth curves were monitored for the B. subtilis wild type (WT), the bshA (A), ohrA (B), hypR (C), and spx mutants (D) before and after treatment with 90 µM allicin stress. The strains were cultivated in BMM and exposed to allicin at an OD 500 of 0.4 All mutants were impaired in growth under allicin stress. The results are from 3 biological replicates. Error bars represent the SD. The statistics was calculated using a Student's unpaired two-tailed t-test by the GraphPad Prism software. The p-values are p < 0.0001 at 120-270 min in (A); p < 0.001 at 120-180 min in (B), p < 0.0001 at 60-210 min in (C), and p < 0.01 at 150-240 min in (D). Symbols are ns p > 0.05, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

Discussion
The antimicrobial and toxic effects of garlic can be attributed to the thiosulfinate allicin and diallyl polysulfanes of different sulfur chain length, which are generated during damage, heating, or aging of garlic tissues [5]. The antimicrobial mode of action of garlic-derived reactive sulfur compounds is mainly caused by its thiol-reactivity and depletion of the reduced thiol pool, including LMW thiols (GSH, BSH) and protein thiols. Allicin and diallyl polysulfanes react with LMW thiols via S-thioallylation, as demonstrated in the thiol-metabolomes of E. coli, B. subtilis, S. aureus, yeast, and human Jurkat cells [5,7,9,19,20,26]. However, the comparative mode of actions of allicin and polysulfanes as degradation products have not been studied in bacteria. It was further interesting to investigate whether allicin and polysulfanes modify similar protein targets by S-thioallylation. Thus, we used global transcriptomics and proteomics approaches to investigate in more detail the mode of action and targets for S-thioallylation in B. subtilis by both allicin and DAS4 in comparison. Using pulldown assays, we identified S-thioallylations at the redox-sensing Cys residues of the MarR-type repressors OhrR, HypR, CatR, and YodB as redox-sensing mechanisms under allicin stress leading to up-regulation of the corresponding redox regulons in the transcriptome. Phenotype analyses of mutants revealed potential roles of BSH, OhrA, HypR, and Spx in protection against allicin toxicity in B. subtilis. These results revealed new roles for the thiol-specific OhrA, HypR, and Spx regulons in protection under allicin stress.
Using microarrays, we could show that allicin and DAS4 cause a similar transcriptome signature, which is indicative of a strong oxidative, disulfide, and sulfur stress response in B. subtilis. The transcriptome comparison between the allicin and DAS4 responses in B. subtilis revealed similar high inductions of most thiol-stress response regulons, such as OhrR, PerR, HypR, Spx, CatR, YodB, and AdhR. In addition, allicin and DAS4 provoked a strong metal stress response and protein unfolding due to disulfide formation as shown by the induction of the ArsR, CzrA, CsoR, and CtsR regulons. Due to related transcriptome signatures, we can conclude that allicin and DAS4 show similar thiol-reactivities in B. subtilis. The allicin and DAS4 thiol-stress signatures overlap strongly with the transcriptome response of B. subtilis under HOCl stress, which causes S-bacillithiolation in the proteome [27,30,31]. In addition, allicin caused strong inductions of the related redox regulons controlled by HypR, PerR, QsrR, CsoR, and CstR in the RNAseq transcriptome of S. aureus [19], showing similar disulfide and metal stress responses as well as protein damage by allicin in both bacteria.
However, while the HypR-controlled flavin disulfide reductase merA gene was most strongly up-regulated under allicin and HOCl in S. aureus [19,58], the top scorer was the OhrR-regulated ohrA gene in B. subtilis under allicin, DAS4 and HOCl stress [27]. In S. aureus, we used NADPH coupled assays in vitro and phenotype analyses of the merA mutant in vivo to reveal an important function of MerA as allicin reductase as novel mechanism for allicin detoxification [19]. Since the ohrA mutant was impaired in growth under allicin stress, the peroxiredoxin OhrA could play a similar role in allicin reduction in B. subtilis. Moreover, the Spx and HypR regulators of B. subtilis control several putative thiol-disulfide oxidoreductases and flavoenzymes, which are highly up-regulated in the transcriptome (e.g., nfrA, yugJ, ywcH, yjbH, and hypO). In addition, hypR and spx mutants showed strong sensitivity towards allicin stress. These results suggest that HypR-and Spx-controlled oxidoreductases could also function in allicin detoxification, which remains to be investigated.
The decomposition of allicin and polysulfanes results in the formation of other RSS, including allyl thiols, allyl persulfides, and H 2 S [3,5,62]. Thus, the RSS-responsive CstR regulon was highly induced in the allicin transcriptome of S. aureus [19], including the cstAB operon that encodes for a thiosulfate sulfurtransferase and persulfide dioxygenase-sulfurtransferase [65][66][67][68][69]. In B. subtilis, no related RSS-specific detoxification mechanism is known. Interestingly, the yrkEFHIJ operon was highly induced by allicin and DAS4 in B. subtilis encoding for a sulfur transferase (yrkE) and TusA-like sulfur carrier proteins (yrkFJ) which could function in RSS detoxification. Overall, the transcriptome signatures of allicin and DAS4 indicate that both garlic compounds cause strong thiol-specific oxidative, metal, and sulfur stress responses, as well as protein damage in B. subtilis and S. aureus [19].
The protein damage response by allicin and DAS4 is caused by their S-thioallylation of protein thiols and LMW thiols, leading to depletion of the total cellular thiol pool. Widespread protein S-thioallylations have been mapped in previous studies and this study using shotgun proteomics in the E. coli, S. aureus, B. subtilis, and human Jurkat cell proteomes [7,19,26]. S-thioallylation of proteins leads to loss of protein functions as shown for selected targets, such as GapDH, cysteine protease papain, alcohol dehydrogenases, enolase, and isocitrate lyase in vitro [7,8,19,26].
In the S. aureus allicin proteome, 57 proteins were modified by S-thioallylations, which include mostly abundant proteins involved in protein translation (EF-Tu, EF-Ts, RpsB, RpmG2), biosynthetic enzymes for nucleotides, and amino acids (GlnA, AldA, GuaB) and antioxidant enzymes (KatA, Tpx, BrxA, MsrB) [19]. Among these, 37 proteins were modified at their redox-sensing active sites or conserved Cys residues indicating that garlic compounds target mainly redox-sensitive thiols [19]. In the B. subtilis allicin and DAS4 proteome, 108 S-thioallylated proteins were identified with similar functions in protein translation, the biosynthesis of nucleotides and amino acids, and detoxification of reactive oxygen species (ROS). These targets for S-thioallylations included 44 proteins that were modified at conserved Cys residues. Some redox-sensitive proteins are conserved targets for S-thioallylation by allicin in B. subtilis and S. aureus, such as Ef-Tu, EF-Ts, GlnA, GuaB, PpaC, MsrB, and GltX [19].
In addition, the targets for S-thioallylations under allicin stress overlap with those proteins S-bacillithiolated under HOCl stress in B. subtilis and S. aureus. These data support that garlic compounds target mostly redox-sensitive active sites [19,27,30,31,35,70,71]. The conserved S-thiolated proteins by HOCl and garlic disulfides are mainly metabolic enzymes, such as MetE, YxjG, GuaB, AldA, PpaC, and SerA, that are most likely redox-regulated and protected by S-thioallylation at their active site Cys residues [31]. In support of this, several redox-sensitive MarR-type regulators are among the main targets for S-thioallylation in both bacteria, such as SarA, MgrA, SarR, and SarH1 in S. aureus [19,72] and OhrR, HypR, CatR, and YodB in B. subtilis. Most of these MarR-type regulators were S-thioallylated at their redox-sensing Cys, leading to repressor inactivation (OhrR, CatR, YodB) or activation of HypR and up-regulation of the corresponding regulons. The redox-regulation of OhrR, HypR, CatR, and YodB due to S-thioallylation and induction of the corresponding regulons in the transcriptome was demonstrated in this study.

Conclusions
Altogether, our study revealed that allicin and DAS4 cause widespread S-thioallylation of many abundant proteins with conserved Cys residues, but also lower abundant redox-sensitive transcriptional regulators. Thus, S-thioallylation functions in thiol-protection and redox-regulation in bacteria to facilitate the recovery of growth and survival by induction of detoxification pathways. Our transcriptome and proteome studies further revealed that both allicin and DAS4 induce similar thiol-specific stress responses and hence exert similar thiol-reactivities on redox-sensitive Cys residues. Thus, allicin and polysulfanes can be applied as efficient thiol-reactive antimicrobials.
Supplementary Materials: The following files are available online at http://www.mdpi.com/2076-3921/8/12/605/s1, Table S1: Transcriptome analysis of B. subtilis 168 wild type after Allicin and DAS4 treatment using microarrays, Table S2: Transcriptome analysis of B. subtilis 168 after exposure to Allicin and DAS4 stress and subsets of gene expression changes classified into regulons, Table S3: Identification of 108 S-thioallylated proteins in the B. subtilis 168 allicin and DAS4 proteome using shotgun LC-MS/MS analysis including their spectral counts, Sequest Xcorrs, deltaCn scores and mass deviations, Table S4: Functional classification based on TIGRfam of the 108 proteins with S-thioallylated Cys peptides identified in the B. subtilis proteome under allicin and DAS4 stress, Table S5: Total spectral counts of all identified proteins in the B. subtilis proteome under allicin an DAS4 stress, Table S6: Identification of 108 S-thioallylated proteins in the proteome of B. subtilis 168 under allicin and DAS4 stress using shotgun LC-MS/MS analysis and the spectral counts of identified S-thioallylated Cys peptides. Funding: This work was funded by an European Research Council (ERC) Consolidator grant (GA 615585) MYCOTHIOLOME and grants from the Deutsche Forschungsgemeinschaft, Germany (AN746/4-1 and AN746/4-2) within the SPP1710 on "Thiol-based Redox switches", by the SFB973 (project C08) and TR84 (project B06) to H.A. Further support for allicin synthesis was provided by internal funding from the RWTH Aachen University to M.C.H.G. and A.J.S. The synthesis and purification of DAS4 was funded by EU FP-7 ITN grant number 215009 and BBSRC grant number BB/N004817/1 to C.J.H.