JOURNAL OF BACTERIOLOGY, Mar. 2008, p. 1751–1761 Vol. 190, No. 50021-9193/08/$08.00�0 doi:10.1128/JB.01516-07Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Transcriptome Analysis of Sorbic Acid-Stressed Bacillus subtilisReveals a Nutrient Limitation Response and Indicates Plasma

Membrane Remodeling�†Alex Ter Beek,* Bart J. F. Keijser,‡ Andre Boorsma,§ Anna Zakrzewska,¶ Rick Orij,

Gertien J. Smits, and Stanley BrulLaboratory for Molecular Biology and Microbial Food Safety, Swammerdam Institute for Life Sciences,

University of Amsterdam, Amsterdam, The Netherlands

Received 19 September 2007/Accepted 13 December 2007

The weak organic acid sorbic acid is a commonly used food preservative, as it inhibits the growth of bacteria,yeasts, and molds. We have used genome-wide transcriptional profiling of Bacillus subtilis cells during mildsorbic acid stress to reveal the growth-inhibitory activity of this preservative and to identify potential resistancemechanisms. Our analysis demonstrated that sorbic acid-stressed cells induce responses normally seen uponnutrient limitation. This is indicated by the strong derepression of the CcpA, CodY, and Fur regulon and theinduction of tricarboxylic acid cycle genes, SigL- and SigH-mediated genes, and the stringent response.Intriguingly, these conditions did not lead to the activation of sporulation, competence, or the general stressresponse. The fatty acid biosynthesis (fab) genes and BkdR-regulated genes are upregulated, which mayindicate plasma membrane remodeling. This was further supported by the reduced sensitivity toward the fabinhibitor cerulenin upon sorbic acid stress. We are the first to present a comprehensive analysis of thetranscriptional response of B. subtilis to sorbic acid stress.

The food industry commonly utilizes sorbic acid and otherweak organic acids as preservatives. Sorbic acid (trans-trans-2,4-hexadienoic acid) is a six-carbon unsaturated fatty acid witha pKa of 4.76 and was first isolated from unripe berries ofRowan (Sorbus aucuparia). The acid, or its anionic salt, is usedin a variety of food products and has a broad range of antimi-crobial activities against spoilage bacteria, yeasts, and molds (5,17, 57). However, the exact mechanism by which sorbate in-hibits microbial growth is not entirely understood. No singlemechanism appears to explain its toxicity to various spoilageorganisms.

Depending on the pKa of the acid and the pH of the envi-ronment, in solution, sorbate exists in equilibrium between thedissociated state (S�) and the undissociated state (HS). Theneutral HS is lipid permeable and able to diffuse into the cell,reaching an equilibrium when the inside and outside concen-trations of HS are equal. Inside, a new equilibrium is formedbetween S� and HS, releasing protons into the cytosol. Thismay acidify the cytosol, causing an inhibition of many meta-

bolic functions (11, 59). Furthermore, the lipophilic tail of thesorbate molecule has been shown to disrupt the membrane andinterfere with membrane proteins (65). This, together with theentry of protons, could result in a loss of the proton motiveforce, disrupting oxidative phosphorylation and affecting thetransport of nutrients (4, 27, 58). Also, the accumulation of S�

in the cell could cause a rise in osmolarity and affect cytosolicenzymes (3, 77).

In order to counteract the effects of sorbic acid, microorgan-isms use various resistance mechanisms. Saccharomyces cerevi-siae uses H�-ATPases to pump out the excess protons at thecost of ATP to maintain pH homeostasis (36, 45) and inducesa dedicated ATP binding cassette (ABC) transporter, Pdr12, toprevent the accumulation of the anion S� (35). These pro-cesses, however, may reduce energy resources significantly (8,35, 36). Studies with benzoic acid showed that adapted S.cerevisiae and Zygosaccharomyces bailii cells reduce their per-meabilities to benzoate (33, 71). Changes in fatty acid compo-sition in sorbate-stressed Zygosaccharomyces rouxii cells havebeen reported (29). Z. bailii is able to degrade benzoate andsorbate (51), and species of Penicillium can decarboxylate sor-bate to 1,3-pentadiene (42).

Compared to yeasts, very little is known about specific weak-acid resistance mechanisms in bacteria. Depending on the spe-cies, bacteria can induce several systems to counteract a dropin the internal pH when encountering low pH stress. Amongothers, these include proton pumps, several decarboxylases(lysine, glutamate, and arginine), the production of urease,arginine deiminase, chaperones (e.g., DnaK and GroELS), andsigma factor (SigB, SigM, and RpoS)-mediated responses (5, 6,14, 31, 67). However, the importance of low-pH stress responsesystems in weak-acid resistance development remains unclear.

The gram-positive bacterium Bacillus subtilis is one of the

* Corresponding author. Mailing address: Laboratory for MolecularBiology and Microbial Food Safety, Swammerdam Institute for LifeSciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018WV Amsterdam, The Netherlands. Phone: 31 20 5257025. Fax: 31 205257056. E-mail: aterbeek@science.uva.nl.

† Supplemental material for this article may be found at http://jb.asm.org/.

‡ Present address: TNO Quality of Life, Food and BiotechnologyInnovations—Microbiology, Utrechtseweg 48, 3704 HE Zeist, TheNetherlands.

§ Present address: TNO Quality of Life, Physiological Genomics,Utrechtseweg 48, 3704 HE Zeist, The Netherlands.

¶ Present address: Institute of Biology, Leiden University, Was-senaarseweg 64, 2333 AL Leiden, The Netherlands.

� Published ahead of print on 21 December 2007.

1751

organisms that causes food spoilage, and its growth is inhibitedby sorbic acid (25). This rod-shaped bacterium commonly livesin the upper layers of soil and is therefore found on crops andin food products. Thus, we investigated the time-resolved ge-nome-wide response of B. subtilis sublethally stressed with po-tassium sorbate (KS) using DNA microarray technology. Weused the complementary methods of hierarchical clusteringand T-profiler, adapted for B. subtilis, to analyze the data. Ourresults indicate that sorbic acid induces responses normallyseen upon nutrient limitation. However, mild sorbic acid stressdoes not lead to the induction of the general stress response(GSR), sporulation, or competence. B. subtilis likely remodelsits plasma membrane, possibly to reduce the entry of sorbicacid into the cell.

MATERIALS AND METHODS

Bacterial strains and growth conditions. All B. subtilis strains used in thisstudy are derivatives of the laboratory wild-type (WT) strain PB2 (trp2C). PB2,PB153 (trpC2 sigB�2::cat), and PB198 (amyE::Pctc-lacZ) were kindly provided byC. W. Price. Mutant strains ATB002 (ureC), ATB008 (sigL), and ATB003 (padC)were obtained by transformation of strain PB2 with chromosomal DNA of strainsSF168U (trpC2 ureC::spc) (15), QB5505 (trpC2 sigL::aphA3) (19), and BS783(trpC2 padC::cat) (23), respectively. To obtain mutant strains ATB001 (fabHB),ATB004 (ycsF), ATB005 (yhcA), ATB006 (yhcB), and ATB007 (yxkJ), WT strainPB2 was transformed with chromosomal DNA of strains YHFBd (trpC2 fabHB::pMUTIN), YCSFd (trpC2 ycsF::pMUTIN), YHCAd (trpC2 yhcA::pMUTIN),YHCBd (trpC2 yhcB::pMUTIN), and YXKJd (trpC2 yxkJ::pMUTIN), respec-tively, which were all received from the Japanese Consortium for FunctionalAnalysis of the B. subtilis Genome (http://bacillus.genome.ad.jp/). Transformantswere selected on Luria-Bertani (LB) agar plates containing appropriate antibi-otics after overnight incubation at 37°C. Depending on the strain, the antibioticsused were chloramphenicol (6 �g/ml), erythromycin (0.5 �g/ml), spectinomycin(100 �g/ml), or kanamycin (10 �g/ml). Isolation of chromosomal DNA wasperformed according to methods described previously by Ward and Zahler (70),and transformations were carried out as described previously by Kunst andRapoport (44).

B. subtilis strains were cultivated in a defined minimal medium as describedpreviously by Neidhardt et al. (56), as modified by Hu et al. (37). The mediumwas buffered with 80 mM 3-(N-morpholino)propanesulfonic acid (MOPS), andthe pH was set to 5.9, 6.4, 7.4, or 7.8 with KOH. As carbon and nitrogen sources,5 mM glucose, 10 mM glutamate, and 10 mM NH4Cl were used. A fivefold (25mM) increase in glucose or a 10-fold (100 �M) or a 25-fold (250 �M) increasein iron was used where indicated. All strains were grown exponentially, trans-ferred into a SpectroMax Plus microtiter plate reader (Molecular Devices Corp.)at an optical density at 600 nm (OD600) of 0.08 (which corresponds to an OD600

of 0.2 in a 1-cm-path-length spectrophotometer), and stressed with various con-centrations of KS ranging from 1.25 to 125 mM or 5 �g/ml cerulenin whereindicated. Cells were further cultivated in the microtiter plate reader underrigorous shaking at 37°C for 180 min. All conditions were tested in the microtiterplate reader at least in duplicate, and biologically independent experiments wereperformed at least twice.

Assay of �-galactosidase activity. PB198 (amyE Pctc-lacZ) was grown expo-nentially in shake flasks in defined medium at pH 6.4 to an OD600 of 0.2 andstressed with 3, 7, and 20 mM KS or 0.3 M NaCl. To determine the �-galacto-sidase activity, 1-ml samples were collected every 15 min for 1 h, frozen in liquidnitrogen, and stored at �20°C until further processing. The �-galactosidase assaywas performed as described previously (41). Cells were permeabilized using0.002% sodium dodecyl sulfate and 4% chloroform (final concentrations). LacZactivities were calculated as Miller units (48).

Preparation of total RNA for transcriptome analysis and real-time reversetranscriptase (RT) PCR. An exponentially growing culture of B. subtilis WTstrain PB2 was split into two cultures and inoculated in well-controlled batchfermentors (500-ml working volume) to an OD600 of 0.05. The cultures weregrown at 37°C in defined medium at pH 6.4 with an aeration rate of 0.5 liters/minand vigorous stirring (200 rpm). At an OD600 of 0.2, one culture was stressed with3 mM KS. Samples of 20 ml were withdrawn from both the treated and controlcultures at 0, 10, 20, 30, 40, and 50 min after the addition of KS. Glucose levelsand oxygen consumptions were obtained as described elsewhere previously (1).The cells were collected using a vacuum-filtering setup, immediately quenched in

liquid nitrogen, and stored at �80°C prior to RNA extraction. The whole pro-cedure took no longer than 50 s. Two biologically independent experiments wereperformed. Total RNA was isolated as described previously (40).

Synthesis of labeled cDNA, hybridization, and scanning of the DNA microar-rays. Superscript II RT (Invitrogen) was used to synthesize labeled cDNA fromtotal RNA samples by the direct incorporation of Cy3- or Cy5-labeled dUTP intocDNA. The reaction mixture in first-strand buffer contained 12 �g of total RNA;0.5 �g of random hexamers (GE Healthcare); 400 units of Superscript II RT; 10mM dithiothreitol; 0.5 mM dATP, dCTP, and dGTP; 0.2 mM dTTP (GE Health-care); and 0.07 mM Cy3- or Cy5-dUTP (GE Healthcare). Control and sorbate-treated samples were incorporated with Cy3- and Cy5-labeled dUTP, respec-tively. After cDNA synthesis, the RNA was hydrolyzed using 1.5 �l of 1 M NaOHfor 10 min at 70°C. The pH was neutralized with 1.5 �l of 1 M HCl, and thelabeled cDNA was purified by using QIAquick PCR purification spin columns(Qiagen). The efficiency of labeling was monitored spectrometrically on a Nano-drop apparatus (Isogen Life Science). The B. subtilis DNA microarrays wereconstructed as described previously by Keijser et al. (40). Each constructed arraycontained spots in duplicate with 4,100 gene-specific 65-mer oligonucleotidesrepresenting 4,100 of the 4,106 protein-coding genes in B. subtilis (as reported forthe B. subtilis genome at http://genolist.pasteur.fr/SubtiList/). Hybridization andscanning was performed as described previously (40).

Microarray data extraction and processing. Quantification of the hybridiza-tion signals from both Cy3 and Cy5 channels and background subtractions werecarried out with ArrayVision 6.1 software (Imaging Research Inc.). First, thepixels with density values that exceeded four median absolute deviations abovethe median were removed, and the average of all pixels remaining in the spot wascomputed for each channel (the artifact-removed [ARM] density values). Sec-ond, the local background was calculated for each spot and subtracted from itsARM density value (resulting in the subtracted ARM density value). Spots witha signal-to-noise ratio (subtracted ARM divided by the standard deviation of thelocal background) smaller than 2.0 in both channels were excluded from furtheranalysis. The remaining data were normalized in J-Express Pro 2.6 software(MolMine AS) using a global locally weighted scatterspot-smoothing algorithm(75). To avoid extreme intensity ratios, low-intensity fluorescence data werefloored at a value corresponding to a signal-to-noise ratio of 2.0. The data wereaveraged and log2 transformed, and missing values were replaced by the averageof the closest values. Genes with more than two missing values in the time serieswere omitted. Since the variation in differential expression measurements de-pends on the fluorescent signal intensity (smaller variation at higher fluorescenceintensity levels and larger variation at lower fluorescence intensity levels), weapplied an intensity-dependent method to identify differentially expressed genes(74). A sliding window of 50 genes was selected to calculate a Z score from thelocal mean and standard deviation using the data in the R-I plot [log10(Cy5 �Cy3) versus log10(Cy5/Cy3)]. Genes more than 1.96 standard deviations awayfrom the local average (Z � 1.96) were considered to be differentially ex-pressed. This corresponds to a confidence level of 95%. Genes that showedsignificant expression at 0 min were excluded from further analysis, unless thegene showed the opposite significant expression in at least one of the other timepoints. After the processing of the microarray data, 3,909 genes remained foreach time point, 459 of which were found to be significantly expressed. Thedegree of enrichment or depletion for a specific gene group in the given signif-icantly up- or downregulated genes was quantitatively assessed using a hyper-geometric distribution analysis (55). Gene groups were considered to be enrichedor depleted when the calculated P value was below 0.01.

Microarray data analysis. Hierarchical clustering (24) of the significantlyregulated genes was used to identify groups of genes with similar transcriptionprofiles. In J-Express Pro 2.6 (MolMine AS), all 459 genes showing significantexpression during KS treatment were hierarchically clustered using the averagelinkage (weighted-pair group method using average linkages) clustering methodand a Euclidian distance metric.

To assess the contribution of the expression of genes from specific gene classesto the total gene expression of all 3,909 genes, we used T-profiler (7). T-profilerwas adapted for the use of B. subtilis transcriptome data by implementing pre-defined gene groups from the following sources: the Database of TranscriptionalRegulation in Bacillus subtilis (DBTBS) (release May 2006) (47), the KyotoEncyclopedia of Genes and Genomes database (KEGG) (release May 2006)(39), the SubtiList database (54), and the stringently controlled genes (26). Genegroups regulated positively or negatively by a specific transcription factor werenamed accordingly. For transcription factors acting both as an activator and as arepressor, separate gene groups were made. The composition of the individualgene groups can be found in the above-mentioned sources.

Relative quantification of gene expression using real-time RT-PCR. For real-time RT-PCR, RNA isolated for the DNA microarray was used to make cDNA.

1752 TER BEEK ET AL. J. BACTERIOL.

RT reactions were performed using Superscript II RT (Invitrogen) according tothe manufacturer’s instructions. Equal amounts of total RNA (5 �g) and 150 ngof random hexamers (GE Healthcare) were used in RT reactions. The amplifi-cation and detection of PCR product were performed with the 7300 real-timePCR system (Applied Biosystems). Primer Express 3.0 software (Applied Bio-systems) was used to design specific primers (purchased from Isogen Life Sci-ence) for real-time PCR (see Table S1 in the supplemental material). Reactionswere carried out in a 20-�l mixture consisting of 3 to 9 �M specific primers, 2 �lof 200-fold-diluted cDNA template, and SYBR green PCR master mix (AppliedBiosystems). The cycling conditions were as follows: 1 cycle at 50°C for 2 min, 1cycle at 95°C for 10 min, and 40 cycles at 95°C for 15 s and at 60°C for 1 min.Melting curves were used to monitor the specificity of the reaction. RNA of alltime points and independent experiments used in the microarray analysis wereanalyzed with real-time PCR in duplicate. Because the amplification of the targetand reference genes was tested and found to be approximately equal (notshown), the ��CT method could be used to calculate relative gene expressions(46). The expression levels of the investigated genes were determined relative tothe untreated reference group. The ratios (2���CT) were calculated and log2

transformed. The accA gene was used as the internal control, since the expres-sion of this gene was constant under both control and stress conditions in bothmicroarray and real-time PCR experiments.

Microarray data accession number. Microarray data were deposited in theGEO database (http://www.ncbi.nlm.nih.gov/geo/) under accession numberGSE9823.

RESULTS

Growth inhibition by sorbic acid can be attributed mainly tothe undissociated form of the acid. Weak acids in solution arein equilibrium between their undissociated and dissociatedform. To investigate whether undissociated (HS) or dissociated(S�) sorbic acid is responsible for growth inhibition of B.subtilis, the pH dependence of sorbate action was tested onexponentially growing cells in a defined minimal medium. Byusing a defined and buffered medium, the pH remained stable,and the unwanted presence of weak organic acids in undefinedrich media such as LB broth was avoided. B. subtilis WT strainPB2 was grown exponentially in the presence of KS (0 to 40mM) at pH 5.9, 6.4, 7.4, and 7.8 (Fig. 1A). Clearly, the reduc-tion in the growth rate is KS concentration dependent. Notethat the drop in the OD600 observed at the end of the expo-nential phase coincides with the depletion of glucose (con-firmed by high-performance liquid chromatography) (our un-published data), which may be due to a diffusion limitation ofoxygen in the microtiter plate experiments. Cells grown intostationary phase in the presence of excess glucose did not showthis phenomenon (our unpublished data). Similar qualitativetrends of the growth curves were observed for all pH valuestested. However, at a higher pH, higher concentrations of KSwere needed to similarly affect growth (our unpublished data).We plotted the percentages of growth inhibition against theconcentrations of S� and HS and observed that the curves forHS had a clear overlap, in contrast to those for S� (Fig. 1B andC). This demonstrates that HS is largely responsible for thegrowth inhibition and that its effect is pH independent. How-ever, a closer look at the lower concentration range of HS (Fig.1C, inset) reveals a pH dependence, suggesting that S� alsocontributes to the growth inhibition although to a much lesserextent. These findings are in general agreement with previouslyreported observations (17, 25).

It is noteworthy that sorbic acid stress lowers the maximallyobtained OD600 (Fig. 1A), which suggests a lower yield. Inorder to investigate this in more detail, we grew cells in well-aerated batch fermentors, stressed them with 3 mM potassium

FIG. 1. Growth inhibition of exponentially growing B. subtilis by KSat various pH values. (A) The growth of WT PB2 in defined minimalmedium at pH 6.4 was monitored in a microtiter plate reader. Theclosed circles indicate the growth of the control experiment (no addi-tion of KS). Stress conditions were 2.5 mM (open circles), 5 mM(closed triangles), 10 mM (open triangles), 20 mM (closed squares),and 40 mM (open squares) KS. The OD600 was monitored during 180min. The values represent the means of four measurements, includingthe standard errors. (B and C) Percentage of growth inhibition com-pared to the control (no addition of KS) as a function of the calculatedconcentration of S� (B) and HS (C) molecules. The percentages ofgrowth inhibition were calculated from the increase in the opticaldensity between 56 and 108 min after sorbate stress. The inset shows adetailed part of the lower range of the calculated HS concentration.Experiments were performed at pH 5.9 (closed circles), pH 6.4 (opencircles), pH 7.4 (closed triangles), and pH 7.8 (open triangles). Thevalues represent the means of four measurements, including the stan-dard errors.

VOL. 190, 2008 B. SUBTILIS SORBIC ACID STRESS RESPONSE 1753

sorbate at pH 6.4, and measured the oxygen consumption andglucose levels. This treatment resulted in a 29% reduction inthe growth rate. The values for the calculated yield on oxygendecreased 20%, and those for the calculated yield on glucosedecreased 16% during 50 min of sorbic acid treatment. Thissuggests that sorbate lowers the energetic efficiency of respi-ration, which may be a consequence of a decreased protongradient and/or cellular reprogramming. Compared to the un-treated control, the glucose flux per cell calculated over thistime interval decreased 10%, which indicates that the uptakeof glucose and/or its further metabolism is hampered or ad-justed by the cell.

Time-resolved transcriptome analysis of sorbic acid-treatedB. subtilis. To obtain a better understanding of the response ofB. subtilis enduring sorbic acid stress, DNA microarray analysiswas performed. We studied the changes in gene expression incells exponentially grown in batch fermentors. Samples weretaken at 10, 20, 30, 40, and 50 min after exposure to 3 mM KSat pH 6.4 (29% growth inhibition) and compared to an un-treated control. Using a fluorescent signal intensity-dependentmethod (see Materials and Methods), we identified a total of459 genes (11.2% of the genome) that were differentially ex-pressed in at least one time point in comparison with theuntreated samples (see Table S2 in the supplemental mate-rial). We used real-time RT-PCR to validate the results of themicroarray and selected seven transcripts representative of thevarious temporal expression patterns observed (see Table S3 inthe supplemental material).

Microarray interpretation using hierarchical clustering andT-profiler. To analyze the data obtained from the microarrayexperiments, we used two complementary methods: hierarchi-cal clustering (24), to identify groups of genes (clusters) withsimilar transcription profiles, and T-profiler (7), to determinesignificantly regulated gene groups. All 459 genes that showedsignificant changes were hierarchically clustered and dividedinto four main clusters using an arbitrary distance cutoff, indi-cated by the dashed blue line in the hierarchical tree (Fig. 2).Subdivisions of these clusters are indicated by colored bars andletters next to each cluster. The complete hierarchical cluster-ing, together with all gene names, their descriptions, and func-tional categories, is available in the supplemental material (seeFig. S1 in the supplemental material).

T-profiler (http://www.science.uva.nl/�boorsma/t-profiler-bacillusnew/), developed originally for the analysis of S. cer-evisiae (78) and Candida albicans transcriptome data, wasadapted for the analysis of genome-wide expression data for B.subtilis (see Materials and Methods). T-profiler optimally usesall data, in contrast to hierarchical clustering, where a cutofffor significance is applied. Importantly, T-profiler transformstranscriptional data of single genes into the behavior of genegroups, reflecting biological processes in cells. All gene groupswith significant T values in any time point are presented in thesupplemental material (see Tables S4 to S8 in the supplemen-tal material).

Global adaptive responses to sorbic acid stress in B. subtilis.B. subtilis has different global adaptive responses that can beinduced when it encounters stress or starvation. The GSR,regulated by the sigma factor SigB, is induced by many differ-ent types of stress (e.g., glucose starvation, heat, low externalpH, salt, and ethanol) and provides the cell with nonspecific,

FIG. 2. Hierarchical clustering of significantly regulated genes. All459 genes showing significant expression during KS treatment werehierarchically clustered using the average linkage (weighted-pair groupmethod using average linkages) clustering method and a Euclidiandistance metric (J-Express PRO 2.6; MolMine AS). Genes with similarlevels of expression are grouped into clusters. Times refer to minutesafter the addition of KS to exponentially growing cells. Log2 ratios aredisplayed colorimetrically, ranging from �4.23 to �4.23. Red indicateshigher transcript levels in stressed cells than in control cells; greenindicates a reduction in mRNA content. Four main clusters (using anarbitrary distance cutoff) are indicated by the dashed blue line in thehierarchical tree. Subdivisions of clusters are indicated by colored barsand letters next to each cluster. The log2 ratios of each gene used arethe averages of two independent experiments.

1754 TER BEEK ET AL. J. BACTERIOL.

multiple, and preventive stress resistance (31). Rather unex-pectedly, we found no evidence for the induction of the GSRin our microarray analysis. Groupwise analysis using T-profileryielded no significant values for the SigB-regulated gene group(see Fig. S2A in the supplemental material), and SigB-regu-lated genes were not overrepresented among the 256 upregu-lated genes (2 out of 95 SigB-regulated genes) (P 0.080). Inaddition, the reporter strain PB198 (amyE::Pctc-lacZ), used tomonitor the induction state of the GSR, showed no induction(see Fig. S2B in the supplemental material). Only very stronginhibition of growth (71%) caused by 20 mM KS resulted inincreased LacZ activity. A sigB mutant strain showed suscep-tibility to KS similar to that of the WT strain for all concen-trations tested in liquid medium (Table 1). Additionally, long-term stress survival, as tested by spotting 10-fold dilution seriesof exponentially growing WT and sigB mutant cells on platescontaining KS, revealed no difference, even at high KS con-centrations (our unpublished data). We conclude that the GSRis not the key response of cells encountering sorbic acid stress.

When nutrients become limiting at the end of exponentialgrowth and cells enter the stationary phase, the genes of theSigH regulon and the genes repressed by the transition-stateregulator AbrB and the early-stationary-phase regulator CodYare activated to adapt to limiting conditions (63, 66). T-profilershowed a clear induction of transition-state and early-station-ary-phase genes (Fig. 3A). Indeed, 6 out of 29 (P 0.0021)genes of the SigH regulon were significantly induced (Fig. 2,clusters 2A, 2E, and 4). Also, the repression of CodY- andAbrB-regulated genes was relieved immediately after treat-ment with sorbic acid (Fig. 3A). This observation is furthersupported by the significant induction of 18 out of 42 (P 0.0001) genes negatively regulated by CodY and the inductionof 15 out of 59 (P 0.0001) genes repressed by AbrB as wellas the downregulation of 6 out of 20 (P 0.0001) genespositively regulated by AbrB. These data indicated that theactivity of AbrB decreased transiently in the first 20 min aftersorbate stress. To corroborate the derepression of genes neg-atively regulated by CodY, we analyzed and confirmed the

FIG. 3. Global adaptive responses to sorbic acid stress in B. subtilis as revealed by T-profiler. (A) Induction of transition-state and early-stationary-phase gene groups. The T values of gene groups regulated negatively (closed circles) and positively (open circles) by AbrB and negativelyby CodY (closed triangles) and by SigH (open triangles) are shown. (B) Induction of the stringent response. The T values of gene groups negatively(closed circles) and positively (open circles) controlled by RelA and involved in translation (closed triangles) and protein synthesis (open triangles)are shown. (C) No induction of competence or sporulation. The T values of gene groups regulated positively by ComK (closed circles), negativelyby KipR (open circles), and negatively (closed triangles) and positively (open triangles) by Spo0A are shown. A distinction is made between groupsof genes that are positively or negatively regulated by the transcription factor mentioned (indicated by Pos. and Neg., respectively). The numberof open reading frames within each group present in the microarray experiments is shown in parentheses. The T values presented are calculatedon the basis of two biologically independent experiments. Shown are gene groups that have at least one significant T value (E 0.05) in the timecourse analyzed. The gene groups regulated negatively by ComK and positively by Spo0A (dashed lines) did not show significant T values in thetime course analyzed.

TABLE 1. Susceptibilities of selected mutant strains to sorbic acid stressa

Strain Relevantgenotype

� (h�1) at KSof 0 mMa

GI % (SD) at KS (mM) of:

3 6 15 30

PB2 0.93 (0.03) 28.7 (3.0) 43.4 (1.5) 63.4 (1.8) 73.3 (1.4)ATB001 fabHB 0.87 (0.04) 29.6 (5.2) 41.8 (2.7) 63.8 (1.1) 73.6 (0.3)ATB002 ureC 0.97 (0.02) 31.5 (2.6) 45.1 (0.9) 61.4 (1.3) 71.9 (1.3)ATB003 padC 0.97 (0.01) 19.4 (0.4) 35.2 (0.9) 63.0 (0.7) 75.3 (1.3)ATB004 ycsF 0.87 (0.03) 29.5 (0.5) 41.6 (1.0) 62.5 (0.5) 72.2 (0.5)ATB005 yhcA 0.99 (0.02) 17.1 (2.4) 28.0 (1.4) 63.2 (3.1) 73.7 (3.9)ATB006 yhcB 0.88 (0.02) 28.0 (0.5) 40.2 (2.4) 61.8 (1.4) 70.7 (4.2)ATB007 yxkJ 0.91 (0.04) 30.2 (1.9) 44.4 (1.9) 67.2 (3.7) 75.6 (2.9)PB153 sigB 0.91 (0.01) 27.5 (1.7) 43.3 (0.6) 63.2 (0.5) 74.3 (0.4)ATB008 sigL 0.89 (0.07) 31.0 (5.3) 41.1 (4.5) 65.7 (1.0) 75.3 (4.7)

a KS was added to exponentially growing cultures at an OD600 of 0.2 at pH 6.4. Specific growth rates (�) were calculated from the increase in the optical densitybetween 56 and 108 min after sorbate stress. Growth inhibition percentages (GI %) were calculated from the growth rates obtained and are relative to the specificgrowth rate in the control condition (no addition of KS). All conditions were tested at least in duplicate, and biologically independent experiments were performedat least twice. Standard deviations are shown in parentheses.

VOL. 190, 2008 B. SUBTILIS SORBIC ACID STRESS RESPONSE 1755

induction of the gene ybgE by real-time RT-PCR (see Table S3in the supplemental material). This gene is known to be reg-ulated only by CodY (52). The induction of transition-stateand early-stationary-phase genes suggests that sorbic acid-stressed cells respond as if they experience nutrient limitation.

In amino acid-, glucose-, or oxygen-limited cells, the (RelA-mediated) stringent response helps to prevent the waste ofscarce nutrients (9, 26). Hierarchical clustering revealed thatonly 8 out of 55 (P 0.017) RelA-dependent positive stringentcontrol genes were significantly induced and that 4 out of 86(P 0.79) RelA-dependent negative stringent control geneswere significantly repressed. However, upon considering allgenes, as is done in T-profiler, significant T values for both(RelA-dependent) positive and negative stringent control genegroups were found (Fig. 3B). The significant downregulation ofgene groups involved in translation and protein synthesis (Fig.3B), as well as purine and pyrimidine synthesis (see Tables S4,S6, and S7 in the supplemental material), further reflects theinduction of the stringent response and the observed reductionin growth.

Nutrient limitation can trigger the onset of sporulation andthe development of competence (22, 64). However, we foundno concerted induction of genes controlled by the key compe-tence regulator ComK (Fig. 3C), of other known regulatorsinvolved in competence, or of the functional category trans-formation/competence (our unpublished data). Sporulationwas also not induced by sorbic acid. We found a clear down-regulation of the KipR-regulated genes (Fig. 2, cluster 3E, andFig. 3C), which may indicate the release of the “brake” on thesporulation-regulatory cascade. KipR is the negative regulatorof the ycsFGI-kipIAR-ycsK operon (69). Inactivation of ycsF,the first gene of this operon, did not lead to an altered suscep-tibility to sorbic acid (Table 1). KipI is an inhibitor of KinA, theprimary kinase in the phosphorelay necessary for the phosphor-ylation of the key transcription factor Spo0A that regulates theinitiation of sporulation (64). However, no induction of genespositively regulated by the sporulation master regulator Spo0Awas observed (Fig. 3C). There was a brief downregulation ofgenes repressed by Spo0A (8 out of 17) (P 0.0001), but thiscan be fully explained by the downregulation of the argininebiosynthesis genes controlled by AbrB and the arginine me-tabolism regulator AhrC (Fig. 2, cluster 3D; see Table S4 in thesupplemental material). Furthermore, we did not observe aninduction of other sporulation gene groups regulated bySpoIIID or SpoVT or belonging to the SigE, SigF, SigG, andSigK regulon (our unpublished data). Compared to the un-treated control, no increased spore counts were detected insorbic acid-stressed cultures (our unpublished data). In con-clusion, we see no signs of the development of competence orinduction of sporulation.

Responses to counteract intracellular acidification. One ofthe most prominent effects of weak acids on the microbial cellmay be cytosolic acidification. We discovered a brief but sig-nificant induction of the genes coding for the class I heat shockproteins GroES and GroEL (Fig. 2, clusters 1A and 2A).Chaperones are known to play a role in the maintenance ofprotein folding at a lowered pH, and the induction of groELupon acid stress in lactobacilli, Streptococcus mutans, Clostrid-ium perfringens, and Listeria monocytogenes was described pre-viously (14).

We investigated cellular functions that may counteract theputative pH drop and maintain pH homeostasis. We found nosignificant induction of ATPases and components of the respi-ratory chain. However, the capacity of the cell’s ATPases andrespiratory chain may well be sufficient to maintain pH ho-meostasis without regulation at the transcription level. We didobserve the altered expression of three out of six genes (P 0.0001) belonging to the Gene Ontology group involved in theregulation of pH (GO:0006885). The following genes weresignificantly downregulated: nhaC (Na�/H� antiporter), yjbQ(similar to the Na�/H� antiporter), and yuiF (similar to hypo-thetical proteins). A highly upregulated gene coding for a pro-ton symporter that transports both citrate and malate was yxkJ(43). The greater-than-12-fold induction of this gene (Fig. 2,cluster 4) might indicate its involvement in sorbic acid toler-ance. However, a strain mutated for yxkJ was not more sensi-tive to sorbic acid than the WT (Table 1). Long-term stresssurvival on plates containing KS did not reveal sensitivity ei-ther (our unpublished data).

Interestingly, we found all three genes of the ureABCoperon, which code for the structural components of urease(15), in cluster 4 with the most upregulated genes (Fig. 2). Theregulation of this operon is complex, as it is regulated by PucR,TnrA, CodY, and GlnR, and expression is dependent on SigAand SigH (10, 72). In addition, the induction of ureA is alsoRelA dependent (26). Thus, the induction of the ureABCoperon may be explained by the derepression of CodY-regu-lated genes and the induction of the stringent response (Fig.3A and B). The upregulation of this operon might be involvedin counteracting the possible acidification of the cytosolthrough the production of basic ammonia, as in some bacterialpathogens, ureases are known to facilitate survival in acidicenvironments (13, 49). However, a urease mutant was not KSsensitive in liquid cultures (Table 1) or solid cultures (ourunpublished data).

Although sorbic acid is not a phenolic acid, we surprisinglyfound strong upregulation of the phenolic acid decarboxylasepadC (12) in cluster 4 (Fig. 2). The genes with unknown func-tion yveF and yveG, which lie upstream of padC in the genome,were also strongly induced and present in this cluster, whichmay indicate the coregulation of these three genes. The inac-tivation of padC did not lead to KS sensitivity. Surprisingly, themutant strain showed a clear resistant phenotype at low KSconcentrations both in liquid cultures (Table 1) and on solidcultures (our unpublished data).

Responses to cope with a decreased proton gradient. Theproposed influx of protons mediated by the diffusion of theweak acid over the cell membrane may lead to a decreasedproton motive force. Also, the expulsion of protons to maintainpH homeostasis by ATPases and/or the respiratory chain couldlead to a higher demand on energy resources, which shouldhave a negative effect on the yield. The latter was indeedmeasured as reported above.

Interestingly, we observed major changes in genes involvedin carbon metabolism upon sorbic acid stress (Fig. 4B). Besidesthe induction of carbohydrate metabolism as well as fructoseand mannose metabolism gene groups, we found a strong up-regulation of genes negatively regulated by the carbon catab-olite control protein CcpA after 20 min of sorbic acid exposure.Carbon catabolite derepression normally occurs when the pre-

1756 TER BEEK ET AL. J. BACTERIOL.

ferred carbon source (glucose, fructose, and mannose) is de-pleted (21). However, glucose was still present in the mediumand consumed by the cells during sorbic acid treatment (con-firmed by high-performance liquid chromatography) (our un-published data), although the glucose flux per cell decreased by10%. Noticeably, a fivefold (25 mM) increase in the glucoseconcentration in medium did not alter the growth-inhibitoryeffect of sorbic acid (Table 2).

The expression of genes dependent on SigL (involved inalternative carbon and nitrogen metabolism) also producedsignificant T values after 20 min of sorbic acid exposure (Fig.4B). The induction of the SigL-dependent levD gene was con-firmed by real-time RT-PCR (see Table S3 in the supplemental

material). However, the inactivation of the sigL gene did notchange the susceptibility to sorbic acid compared to the WT(Table 1).

Genes of the tricarboxylic acid (TCA) cycle were also in-duced gradually during sorbic acid stress (Fig. 4B). The TCAcycle can be fed by additional acetyl coenzyme A producedfrom the breakdown of the catabolic products acetate andacetoin. Indeed, we found a significant induction of the acetylcoenzyme A synthetase acsA, which is responsible for the deg-radation of acetate (30), and acuABC, which is involved in theactivation of AcsA (28). Also, the genes acoA and acoR, in-volved in the breakdown of acetoin (2, 38), were significantlyupregulated.

In addition to these carbon metabolism-related gene func-tions, we observed a short downregulation followed by a strongupregulation of the gene group regulated by the central iron-regulatory protein Fur (ferric uptake repressor) (Fig. 4B). Thistransient profile was also clearly seen in clusters 1D and 2C ofthe hierarchical clustering, containing many Fur-regulatedgenes (Fig. 2). In total, we found 15 out of 36 (P 0.0001)Fur-regulated genes significantly expressed in our analysis withsimilar transient expression profiles. The Fur regulon is nor-mally derepressed under iron-limiting or anaerobic conditions(53, 76). We tested whether increased iron concentrations inthe medium would result in an increase in the stress toleranceagainst sorbic acid. A 10-fold (100 �M) or 25-fold (250 �M)increase of iron in the medium could not decrease the growthinhibition by sorbic acid (Table 2). These combined observa-tions suggest that although nutrients in the cell’s environmentare not limiting, the cells respond as if nutrients are scarce.

Sorbic acid influences the biogenesis of the cell envelope.Since HS can dissolve in the cell membrane and is thought toaffect membrane integrity, we expected to see a membrane-remodeling response or an adaptation of the cell envelope tolimit HS entry. Functions associated with the cell surface ortransport are controlled by extracytoplasmic function sigmafactors, like SigM, SigW, and SigX (32). Indeed, the genegroups regulated by extracytoplasmic function sigma factorsSigW and SigX showed altered expression upon sorbic acidtreatment (Fig. 4A). We also detected a similar repressionpattern for genes regulated by YvrH (Fig. 4A). YvrH is in-volved in the maintenance of the cell surface state throughtranscriptional regulation and controls genes containing a pro-

FIG. 4. T-profiler analysis of the transcriptional response to sorbicacid stress reveals induction of possible specific adaptation mecha-nisms. (A) Responses indicating remodeling of the cell envelope. Tvalues of gene groups regulated positively by BkdR (closed circles),SigW (open circles), SigX (closed triangles), and YvrH (open trian-gles) and of gene groups involved in the metabolism of lipids (closedsquares) are shown. (B) Responses indicating adaptation mechanismsagainst uncoupling of the proton gradient and energy and nutrientlimitation. The T values of gene groups regulated negatively by CcpA(closed circles), Fur (open circles), SigL (closed triangles), carbohy-drate metabolism (open triangles), citrate cycle (TCA cycle) (closedsquares), and fructose and mannose metabolism (open squares) areshown. A distinction is made between groups of genes that are posi-tively or negatively regulated by the transcription factor mentioned(indicated by Pos. and Neg., respectively). The number of open read-ing frames within each group present in the microarray experiments isshown in parentheses. The T values presented were calculated on thebasis of two biologically independent experiments. Shown are genegroups that have at least one significant T value (E 0.05) in the timecourse analyzed.

TABLE 2. Growth medium supplemented with either additionalglucose or iron does not increase tolerance to sorbic acid stressa

Glucose addition inmedium(5 mM)

FeSO4 addition inmedium (10 �M) GI % (SD)b

1� 1� 28.7 (3.0)5� 1� 30.0 (0.2)1� 10� 31.2 (0.4)1� 25� 32.7 (2.3)

a See Materials and Methods for the standard composition of the medium.b KS (3 mM) was added to exponentially growing cultures at an OD600 of 0.2

at pH 6.4. Growth inhibition percentages (GI %) were calculated from thegrowth rates obtained between 56 and 108 min after the addition of KS and arerelative to the specific growth rate in the control condition (no addition of KS).All conditions were tested at least in duplicate, and biologically independentexperiments were performed at least twice. Standard deviations are shown inparentheses.

VOL. 190, 2008 B. SUBTILIS SORBIC ACID STRESS RESPONSE 1757

moter site for SigX and sigX itself (62). The repression ofSigW- and SigX-regulated genes upon sorbic acid stress sug-gests their expression under control conditions. This findingwas corroborated by the absolute fluorescence levels. AlthoughSigW and SigX are activated upon cell wall- and membrane-perturbing stresses, we did not have any indication that ourcontrol cells suffered stress. This was supported by a �max of0.93, no induction of the GSR, and no formation of heat-resistant spores in the control culture (our unpublished re-sults).

T-profiler analysis showed a strong induction of genes in-volved in the metabolism of lipids directly after sorbic acidexposure (Fig. 4A). Most fatty acid biosynthesis (fab) genes,negatively regulated by FapR (YlpC) (61), were significantlyupregulated. The fab initiation genes fabHB and fabD andelongation genes fabF, fabG, and fabI (Fig. 2, clusters 2A, 2E,and 4) showed significant induction, with similar expressionprofiles. The strong upregulation (19-fold) of fabHB was con-firmed using RT-PCR (Table 2). In addition, we observed thesignificant induction of genes regulated by BkdR in both T-profiler analysis (Fig. 4A) and hierarchical clustering (Fig. 2,clusters 2D and E). BkdR is the activator of the bkd operon,consisting of seven genes involved in the synthesis of precursormolecules for branched-chain fatty acids (18). The induction ofboth the fab genes and the bkd operon in sorbic acid-treatedcells will likely increase the number of long-chain andbranched-chain fatty acids in the membrane (20, 61).

We investigated the role of fatty acid synthesis in the resis-tance to sorbic acid stress by growing a �-ketoacyl-acyl carrierprotein (ACP) synthase III mutant in the presence of sorbicacid. Strain ATB001 (fabHB) did not reveal a sensitive pheno-type (Table 1). The antibiotic cerulenin inhibits the �-ketoacyl-ACP synthases (FabHA, FabHB, and FabF) of the fatty acidchain elongation step (16). In addition, cerulenin induces theFapR regulon (60). Remarkably, the simultaneous addition ofsorbic acid and cerulenin to exponentially growing WT cellssignificantly decreased the inhibitory effect of the antibioticcompared to cultures treated with cerulenin alone (Fig. 5).This result indicates that the addition of KS may lead to a

membrane adaptation that alters the cell’s sensitivity to ceru-lenin.

Possible extrusion of the anion. S. cerevisiae uses the pumpPdr12 to extrude the sorbate anion to prevent accumulation(35). Interestingly, we found the gene yhcA to be highly up-regulated in the presence of KS (Fig. 2, cluster 4). YhcA is amultidrug resistance transporter homologue of the major fa-cilitator superfamily. To investigate the role of this gene insorbate stress tolerance, we tested mutant strain ATB005(yhcA) for sorbic acid sensitivity. Surprisingly, the yhcA mutantstrain showed clear sorbate resistance, especially at low KSconcentrations in liquid medium (Table 1) and on solid cul-tures (our unpublished data). yhcB is located downstream ofyhcA in the genome (similar to the trp repressor binding pro-tein). This gene also showed strong upregulation (Fig. 2, clus-ter 4) and was tested for sorbate stress tolerance. Mutant strainATB006 (yhcB) showed no altered sorbic acid susceptibility(Table 1). These data illustrate that the resistance of the yhcAmutant strain is caused by the inactivation of yhcA itself andnot by affected downstream genes.

DISCUSSION

Our study shows that mainly the undissociated form of sor-bic acid (HS) is responsible for the inhibition of growth of B.subtilis (Fig. 1). These data support the model that suggeststhat the neutral form of the acid enters the cell, where itdissociates, possibly acidifying the cytosol and likely contribut-ing to its inhibitory effect (5, 17). Maintaining the internal pHis of crucial importance for the proper functioning of the cell.We observed several responses that may counteract the possi-ble intracellular acidification caused by the dissociation of HS(e.g., the induction of the urease operon and the phenolic aciddecarboxylase padC). However, thus far, we have no indicationthat these responses are functional in HS tolerance. Besidespossible acidification of the cytosol, the entry of protons willlower the proton gradient. Consequently, this will reduce thecell’s efficiency to produce ATP, which is supported by theobserved lower yields on glucose and oxygen upon sorbic acidstress.

The transcriptome analysis clearly revealed that cellsstressed with sorbic acid respond as if they encounter nutrientlimitation. We observed a clear induction of transition-stateand early-stationary-phase genes regulated by AbrB, SigH, andCodY as well as an induction of the stringent response medi-ated by RelA (Fig. 3). The derepression of the carbon catab-olite control genes, regulated by CcpA, major changes in car-bon metabolism, and the induction of the Fur regulon furthersuggest that even though glucose is still abundant in the me-dium, the cells experience nutrient limitation(s) (Fig. 4B). Thismay be caused by effects of the acid on glucose uptake systemsbut may also be the result of an altered intracellular environ-ment and effects on enzymes involved in the subsequent glu-cose-phosphate catabolism. This possibility is supported by thedecreased glucose flux seen in sorbic acid-treated cells. Futurestudies should aim at (short-term) measurements of intracel-lular metabolites and oxygen consumption in the absence andpresence of sorbic acid. The addition of excess glucose or ironto the medium did not change sorbate susceptibility (Table 2),

FIG. 5. Sorbic acid stress may lead to a membrane adaptation thatrenders the cells resistant to cerulenin. Exponentially growing WT PB2cells were grown without (closed circles) or with the addition of 3 mMKS (open circles), 5 �g/ml cerulenin (CL) (closed triangles), and 3 mMKS plus 5 �g/ml CL (open triangles). The OD600 was monitored during180 min. The values represent the means of four measurements, in-cluding the standard deviations.

1758 TER BEEK ET AL. J. BACTERIOL.

which seems to point out that limitations in the medium arenot the cause of the nutrient limitation responses of the cell.

Sorbic acid did not induce sporulation, competence (Fig.3C), or the GSR (see Fig. S2 in the supplemental material).We conclude that the limiting conditions caused by 3 mMsorbic acid are not severe enough to switch on sporulation,competence, or the GSR or that other factors inhibit theseadaptation pathways. Zhang and Haldenwang (79) recentlyshowed that the induction of the GSR by nutritional stress ispreceded by a drop in ATP levels. The absence of a GSRinduction upon (mild) sorbic acid stress might indicate thatATP levels are not severely affected.

The observed active state of the SigW and SigX regulon incontrol B. subtilis cells grown in defined (MOPS-based) me-dium is somehow surprising. Note that we opted for the use ofan established Bacillus medium (see, e.g., references 37 and73) and noted no signs of stress (see Results). The observeddownregulation of SigW- and SigX-regulated genes upon sor-bic acid stress may lead to a blockage of cell envelope remod-eling and, consequently, altered cell envelope composition.

Immediately after sorbic acid exposure, a clear activation ofgenes involved in the metabolism of lipids was observed (Fig.4A). The induction of both the fab genes and the bkd operonwill likely increase the number of long-chain and branched-chain fatty acids in the membrane (20, 61). An elongation offatty acids in the membrane might lower the diffusion rate ofsorbic acid into the cell. The simultaneous addition of morebranched-chain fatty acids into the membrane through theaction of the bkd operon will likely preserve fluidity. The in-activation of fabHB did not result in a sensitive phenotype forsorbic acid (Table 1). Noticeably, since fabHA codes for a�-ketoacyl-ACP synthase III as well, this gene is likely to takeover the function of fabHB in the mutant strain tested. Re-cently, Thomaides et al. (68) illustrated that the fabHA fabHBdouble mutant is nonviable. The combined treatment of cul-tures with both KS and cerulenin, the inhibitor of the �-keto-acyl-ACP synthases, significantly lowered the inhibitory effectof cerulenin alone (Fig. 5), supporting the inferred remodelingof the membrane in response to sorbic acid. Alteration of theplasma membrane composition upon sorbic acid stress in Z.rouxii was described previously (29). In addition, S. cerevisiaeand Z. bailii cells adapted to benzoic acid have a reducedpermeability to benzoate (33, 71). We suggest that in B. subtilis,the observed induction of the fab genes and the bkd operon isa direct response to sorbic acid that will likely alter the mem-brane composition such that the diffusion rate of the organicmolecule over the membrane is lowered.

Accumulation of the anion may also cause harm to the cell(17, 34). The strong upregulation of yhcA, encoding a majorfacilitator superfamily multidrug resistance transporter homo-logue, suggested a potential anion extrusion mechanism. Theinactivation of yhcA, however, revealed a clear resistant phe-notype upon stress with low concentrations of KS in definedminimal liquid medium (Table 1) and on solid cultures (ourunpublished data) under our test conditions. In theory, theextrusion of S� by YhcA may create a futile cycle and depletecellular energy pools, since the anion can reassociate in theextracellular environment with a proton and diffuse back intothe cell, thereby reducing the proton gradient further. If suchfutile cycling indeed occurs, it may significantly contribute to

the observed resistance of the yhcA mutant strain when testedin defined minimal medium. Alternative explanations, in whichthe YhcA protein would be the site at which sorbic acid entersthe cells inadvertently and its deletion would thus lower thesorbic acid sensitivity of the cells, are also possible. Such eventswould be analogous to the recently reported acetic acid stressresistance mechanisms in yeast involving the Fps1 aquaglyc-eroporin protein (50). Interestingly, we have preliminary datashowing that in rich medium, the yhcA mutant strain is moresensitive to KS on solid cultures (our unpublished data). Fu-ture studies should address the function of YhcA and its rela-tionship to the differences observed when minimal or richmedium is used.

In conclusion, sorbic acid induces responses normally seenupon nutrient limitation. Therefore, we suggest that the entryof the protonated acid and the subsequent lowering of theproton gradient increase the demand for energy. The cells donot (and likely cannot) increase the uptake rate of glucose andconsequently experience nutrient limitation. The upregulationof the TCA cycle and the utilization of acetate and acetoin mayprovide sufficient energy to maintain growth but at a lowerrate. Finally, the plasma membrane is remodeled, likely in anattempt to reduce the entry of sorbic acid into the cell.Whether the observed responses in B. subtilis upon sorbic acidstress are representative of weak acids in general remains to beelucidated.

ACKNOWLEDGMENTS

We thank Chester W. Price, Susan H. Fisher, Yasutaro Fujita,Kunio Yamane, Yoshito Sadaie, Haike Antelmann, and Georges Rap-oport for sending strains. We acknowledge Jurgo Verkooijen andTessa Dillerop-van der Hoeven of the Microarray Department of Am-sterdam for the hybridization and scanning of the microarray slides,Muus de Haan for assistance with real-time RT-PCR, and ReubenSmith and Rik van Arnhem for technical assistance. Furthermore, wethank Klaas J. Hellingwerf for critically reading the manuscript andJoost Teixeira de Mattos as well as the anonymous reviewers forthoughtful suggestions.

REFERENCES

1. Alexeeva, S., K. J. Hellingwerf, and M. J. Teixeira de Mattos. 2003. Require-ment of ArcA for redox regulation in Escherichia coli under microaerobic butnot anaerobic or aerobic conditions. J. Bacteriol. 185:204–209.

2. Ali, N. O., J. Bignon, G. Rapoport, and M. Debarbouille. 2001. Regulation ofthe acetoin catabolic pathway is controlled by sigma L in Bacillus subtilis. J.Bacteriol. 183:2497–2504.

3. Azukas, J. J., R. N. Costilow, and H. L. Sadoff. 1961. Inhibition of alcoholicfermentation by sorbic acid. J. Bacteriol. 81:189–194.

4. Bauer, B. E., D. Rossington, M. Mollapour, Y. Mamnun, K. Kuchler, andP. W. Piper. 2003. Weak organic acid stress inhibits aromatic amino aciduptake by yeast, causing a strong influence of amino acid auxotrophies on thephenotypes of membrane transporter mutants. Eur. J. Biochem. 270:3189–3195.

5. Beales, N. 2004. Adaptation of microorganisms to cold temperatures, weakacid preservatives, low pH, and osmotic stress: a review. Compr. Rev. FoodSci. Food Saf. 3:1–20.

6. Bearson, S., B. Bearson, and J. W. Foster. 1997. Acid stress responses inenterobacteria. FEMS Microbiol. Lett. 147:173–180.

7. Boorsma, A., B. C. Foat, D. Vis, F. Klis, and H. J. Bussemaker. 2005.T-profiler: scoring the activity of predefined groups of genes using geneexpression data. Nucleic Acids Res. 33:W592–W595.

8. Bracey, D., C. D. Holyoak, and P. J. Coote. 1998. Comparison of the inhib-itory effect of sorbic acid and amphotericin B on Saccharomyces cerevisiae:is growth inhibition dependent on reduced intracellular pH? J. Appl. Mi-crobiol. 85:1056–1066.

9. Braeken, K., M. Moris, R. Daniels, J. Vanderleyden, and J. Michiels. 2006.New horizons for (p)ppGpp in bacterial and plant physiology. Trends Mi-crobiol. 14:45–54.

10. Brandenburg, J. L., L. V. Wray, Jr., L. Beier, H. Jarmer, H. H. Saxild, and

VOL. 190, 2008 B. SUBTILIS SORBIC ACID STRESS RESPONSE 1759

S. H. Fisher. 2002. Roles of PucR, GlnR, and TnrA in regulating expressionof the Bacillus subtilis ure P3 promoter. J. Bacteriol. 184:6060–6064.

11. Brown, M. H., and I. R. Booth. 1991. Acidulants and low pH, p. 22–43. InN. J. Russell and G. W. Gould (ed.), Food preservatives. Blackie, Glasgow,United Kingdom.

12. Cavin, J. F., V. Dartois, and C. Divies. 1998. Gene cloning, transcriptionalanalysis, purification, and characterization of phenolic acid decarboxylasefrom Bacillus subtilis. Appl. Environ. Microbiol. 64:1466–1471.

13. Collins, C. M., and S. E. D’Orazio. 1993. Bacterial ureases: structure, reg-ulation of expression and role in pathogenesis. Mol. Microbiol. 9:907–913.

14. Cotter, P. D., and C. Hill. 2003. Surviving the acid test: responses of gram-positive bacteria to low pH. Microbiol. Mol. Biol. Rev. 67:429–453.

15. Cruz-Ramos, H., P. Glaser, L. V. Wray, Jr., and S. H. Fisher. 1997. TheBacillus subtilis ureABC operon. J. Bacteriol. 179:3371–3373.

16. D’Agnolo, G., I. S. Rosenfeld, J. Awaya, S. Omura, and P. R. Vagelos. 1973.Inhibition of fatty acid synthesis by the antibiotic cerulenin. Specific inacti-vation of beta-ketoacyl-acyl carrier protein synthetase. Biochim. Biophys.Acta 326:155–156.

17. Davidson, P. M. 2001. Chemical preservatives and natural antimicrobialcompounds, p. 593–627. In M. P. Doyle, L. R. Beuchat, and T. J. Montville(ed.), Food microbiology: fundamentals and frontiers, 2nd ed. ASM Press,Washington, DC.

18. Debarbouille, M., R. Gardan, M. Arnaud, and G. Rapoport. 1999. Role ofbkdR, a transcriptional activator of the sigL-dependent isoleucine and valinedegradation pathway in Bacillus subtilis. J. Bacteriol. 181:2059–2066.

19. Debarbouille, M., I. Martin-Verstraete, F. Kunst, and G. Rapoport. 1991.The Bacillus subtilis sigL gene encodes an equivalent of sigma 54 fromgram-negative bacteria. Proc. Natl. Acad. Sci. USA 88:9092–9096.

20. de Mendoza, D., G. E. Schujman, and P. S. Aguilar. 2002. Biosynthesis andfunction of membrane lipids, p. 43–55. In A. L. Sonenshein, J. A. Hoch, andR. Losick (ed.), Bacillus subtilis and its closest relatives: from genes to cells.ASM Press, Washington, DC.

21. Deutscher, J., A. Galinier, and I. Martin-Verstraete. 2002. Carbohydrateuptake and metabolism, p. 129–150. In A. L. Sonenshein, J. A. Hoch, and R.Losick (ed.), Bacillus subtilis and its closest relatives: from genes to cells.ASM Press, Washington, DC.

22. Dubnau, D. 1991. Genetic competence in Bacillus subtilis. Microbiol. Rev.55:395–424.

23. Duy, N. V., U. Mader, N. P. Tran, J. F. Cavin, L. T. Tam, D. Albrecht, M.Hecker, and H. Antelmann. 2007. The proteome and transcriptome analysisof Bacillus subtilis in response to salicylic acid. Proteomics 7:698–710.

24. Eisen, M. B., P. T. Spellman, P. O. Brown, and D. Botstein. 1998. Clusteranalysis and display of genome-wide expression patterns. Proc. Natl. Acad.Sci. USA 95:14863–14868.

25. Eklund, T. 1983. The antimicrobial effect of dissociated and undissociatedsorbic acid at different pH levels. J. Appl. Bacteriol. 54:383–389.

26. Eymann, C., G. Homuth, C. Scharf, and M. Hecker. 2002. Bacillus subtilisfunctional genomics: global characterization of the stringent response byproteome and transcriptome analysis. J. Bacteriol. 184:2500–2520.

27. Freese, E., C. W. Sheu, and E. Galliers. 1973. Function of lipophilic acids asantimicrobial food additives. Nature 241:321–325.

28. Gardner, J. G., F. J. Grundy, T. M. Henkin, and J. C. Escalante-Semerena.2006. Control of acetyl-coenzyme A synthetase (AcsA) activity by acetyla-tion/deacetylation without NAD� involvement in Bacillus subtilis. J. Bacte-riol. 188:5460–5468.

29. Golden, D. A., L. R. Beuchat, and H. L. Hitchcock. 1994. Changes in fattyacid composition of various lipid components of Zygosaccharomyces rouxiias influenced by solutes, potassium sorbate and incubation temperature. Int.J. Food Microbiol. 21:293–303.

30. Grundy, F. J., D. A. Waters, T. Y. Takova, and T. M. Henkin. 1993. Identi-fication of genes involved in utilization of acetate and acetoin in Bacillussubtilis. Mol. Microbiol. 10:259–271.

31. Hecker, M., and U. Volker. 2001. General stress response of Bacillus subtilisand other bacteria. Adv. Microb. Physiol. 44:35–91.

32. Helmann, J. D. 2002. The extracytoplasmic function (ECF) sigma factors.Adv. Microb. Physiol. 46:47–110.

33. Henriques, M., C. Quintas, and M. C. Loureiro-Dias. 1997. Extrusion ofbenzoic acid in Saccharomyces cerevisiae by an energy-dependent mecha-nism. Microbiology 143:1877–1883.

34. Hirshfield, I. N., S. Terzulli, and C. O’Byrne. 2003. Weak organic acids: apanoply of effects on bacteria. Sci. Prog. 86:245–269.

35. Holyoak, C. D., D. Bracey, P. W. Piper, K. Kuchler, and P. J. Coote. 1999.The Saccharomyces cerevisiae weak-acid-inducible ABC transporter Pdr12transports fluorescein and preservative anions from the cytosol by an energy-dependent mechanism. J. Bacteriol. 181:4644–4652.

36. Holyoak, C. D., M. Stratford, Z. McMullin, M. B. Cole, K. Crimmins, A. J.Brown, and P. J. Coote. 1996. Activity of the plasma membrane H�-ATPaseand optimal glycolytic flux are required for rapid adaptation and growth ofSaccharomyces cerevisiae in the presence of the weak-acid preservative sorbicacid. Appl. Environ. Microbiol. 62:3158–3164.

37. Hu, P., T. Leighton, G. Ishkhanova, and S. Kustu. 1999. Sensing of nitrogen

limitation by Bacillus subtilis: comparison to enteric bacteria. J. Bacteriol.181:5042–5050.

38. Huang, M., F. B. Oppermann-Sanio, and A. Steinbuchel. 1999. Biochemicaland molecular characterization of the Bacillus subtilis acetoin catabolic path-way. J. Bacteriol. 181:3837–3841.

39. Kanehisa, M., S. Goto, M. Hattori, K. F. Aoki-Kinoshita, M. Itoh, S. Ka-washima, T. Katayama, M. Araki, and M. Hirakawa. 2006. From genomicsto chemical genomics: new developments in KEGG. Nucleic Acids Res.34:D354–D357.

40. Keijser, B. J., A. Ter Beek, H. Rauwerda, F. Schuren, R. Montijn, H. van derSpek, and S. Brul. 2007. Analysis of temporal gene expression during Bacil-lus subtilis spore germination and outgrowth. J. Bacteriol. 189:3624–3634.

41. Kenney, T. J., and C. P. Moran, Jr.1987. Organization and regulation of anoperon that encodes a sporulation-essential sigma factor in Bacillus subtilis.J. Bacteriol. 169:3329–3339.

42. Kinderlerer, J. L., and P. V. Hatton. 1990. Fungal metabolites of sorbic acid.Food Addit. Contam. 7:657–669.

43. Krom, B. P., R. Aardema, and J. S. Lolkema. 2001. Bacillus subtilis YxkJ isa secondary transporter of the 2-hydroxycarboxylate transporter family thattransports L-malate and citrate. J. Bacteriol. 183:5862–5869.

44. Kunst, F., and G. Rapoport. 1995. Salt stress is an environmental signalaffecting degradative enzyme synthesis in Bacillus subtilis. J. Bacteriol. 177:2403–2407.

45. Lambert, R. J., and M. Stratford. 1999. Weak-acid preservatives: modellingmicrobial inhibition and response. J. Appl. Microbiol. 86:157–164.

46. Livak, K. J., and T. D. Schmittgen. 2001. Analysis of relative gene expressiondata using real-time quantitative PCR and the 2(�delta delta C(T)) method.Methods 25:402–408.

47. Makita, Y., M. Nakao, N. Ogasawara, and K. Nakai. 2004. DBTBS: databaseof transcriptional regulation in Bacillus subtilis and its contribution to com-parative genomics. Nucleic Acids Res. 32:D75–D77.

48. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring HarborLaboratory Press, Cold Spring Harbor, NY.

49. Mobley, H. L., M. D. Island, and R. P. Hausinger. 1995. Molecular biologyof microbial ureases. Microbiol. Rev. 59:451–480.

50. Mollapour, M., and P. W. Piper. 2007. Hog1 mitogen-activated proteinkinase phosphorylation targets the yeast fps1 aquaglyceroporin for endocy-tosis, thereby rendering cells resistant to acetic acid. Mol. Cell. Biol. 27:6446–6456.

51. Mollapour, M., and P. W. Piper. 2001. The ZbYME2 gene from the foodspoilage yeast Zygosaccharomyces bailii confers not only YME2 functions inSaccharomyces cerevisiae, but also the capacity for catabolism of sorbate andbenzoate, two major weak organic acid preservatives. Mol. Microbiol. 42:919–930.

52. Molle, V., Y. Nakaura, R. P. Shivers, H. Yamaguchi, R. Losick, Y. Fujita, andA. L. Sonenshein. 2003. Additional targets of the Bacillus subtilis globalregulator CodY identified by chromatin immunoprecipitation and genome-wide transcript analysis. J. Bacteriol. 185:1911–1922.

53. Moore, C. M., and J. D. Helmann. 2005. Metal ion homeostasis in Bacillussubtilis. Curr. Opin. Microbiol. 8:188–195.

54. Moszer, I., L. M. Jones, S. Moreira, C. Fabry, and A. Danchin. 2002.SubtiList: the reference database for the Bacillus subtilis genome. NucleicAcids Res. 30:62–65.

55. Motulsky, H. 1995. Intuitive biostatistics. Oxford University Press, NewYork, NY.

56. Neidhardt, F. C., P. L. Bloch, and D. F. Smith. 1974. Culture medium forenterobacteria. J. Bacteriol. 119:736–747.

57. Piper, P., C. O. Calderon, K. Hatzixanthis, and M. Mollapour. 2001. Weakacid adaptation: the stress response that confers yeasts with resistance toorganic acid food preservatives. Microbiology 147:2635–2642.

58. Ronning, I. E., and H. A. Frank. 1987. Growth inhibition of putrefactiveanaerobe 3679 caused by stringent-type response induced by protonophoricactivity of sorbic acid. Appl. Environ. Microbiol. 53:1020–1027.

59. Salmond, C. V., R. G. Kroll, and I. R. Booth. 1984. The effect of foodpreservatives on pH homeostasis in Escherichia coli. J. Gen. Microbiol.130:2845–2850.

60. Schujman, G. E., K. H. Choi, S. Altabe, C. O. Rock, and D. de Mendoza.2001. Response of Bacillus subtilis to cerulenin and acquisition of resistance.J. Bacteriol. 183:3032–3040.

61. Schujman, G. E., L. Paoletti, A. D. Grossman, and D. de Mendoza. 2003.FapR, a bacterial transcription factor involved in global regulation of mem-brane lipid biosynthesis. Dev. Cell 4:663–672.

62. Serizawa, M., K. Kodama, H. Yamamoto, K. Kobayashi, N. Ogasawara, andJ. Sekiguchi. 2005. Functional analysis of the YvrGHb two-component sys-tem of Bacillus subtilis: identification of the regulated genes by DNA mi-croarray and northern blot analyses. Biosci. Biotechnol. Biochem. 69:2155–2169.

63. Sonenshein, A. L. 2005. CodY, a global regulator of stationary phase andvirulence in gram-positive bacteria. Curr. Opin. Microbiol. 8:203–207.

64. Sonenshein, A. L. 2000. Control of sporulation initiation in Bacillus subtilis.Curr. Opin. Microbiol. 3:561–566.

65. Stratford, M., and P. A. Anslow. 1998. Evidence that sorbic acid does not

1760 TER BEEK ET AL. J. BACTERIOL.

inhibit yeast as a classic ‘weak acid preservative. ’ Lett. Appl. Microbiol.27:203–206.

66. Strauch, M. A., and J. A. Hoch. 1993. Transition-state regulators: sentinels ofBacillus subtilis post-exponential gene expression. Mol. Microbiol. 7:337–342.

67. Thackray, P. D., and A. Moir. 2003. SigM, an extracytoplasmic functionsigma factor of Bacillus subtilis, is activated in response to cell wall antibi-otics, ethanol, heat, acid, and superoxide stress. J. Bacteriol. 185:3491–3498.

68. Thomaides, H. B., E. J. Davison, L. Burston, H. Johnson, D. R. Brown, A. C.Hunt, J. Errington, and L. Czaplewski. 2007. Essential bacterial functionsencoded by gene pairs. J. Bacteriol. 189:591–602.

69. Wang, L., R. Grau, M. Perego, and J. A. Hoch. 1997. A novel histidine kinaseinhibitor regulating development in Bacillus subtilis. Genes Dev. 11:2569–2579.

70. Ward, J. B., Jr., and S. A. Zahler. 1973. Genetic studies of leucine biosyn-thesis in Bacillus subtilis. J. Bacteriol. 116:719–726.

71. Warth, A. D. 1989. Transport of benzoic and propanoic acids by Zygosac-charomyces bailii. J. Gen. Microbiol. 135:1383–1390.

72. Wray, L. V., Jr., A. E. Ferson, and S. H. Fisher. 1997. Expression of theBacillus subtilis ureABC operon is controlled by multiple regulatory factorsincluding CodY, GlnR, TnrA, and Spo0H. J. Bacteriol. 179:5494–5501.

73. Wray, L. V., Jr., and S. H. Fisher. 2007. Functional analysis of the carboxy-

terminal region of Bacillus subtilis TnrA, a MerR family protein. J. Bacteriol.189:20–27.

74. Yang, I. V., E. Chen, J. P. Hasseman, W. Liang, B. C. Frank, S. Wang, V.Sharov, A. I. Saeed, J. White, J. Li, N. H. Lee, T. J. Yeatman, and J.Quackenbush. 2002. Within the fold: assessing differential expression mea-sures and reproducibility in microarray assays. Genome Biol. 3:research0062.

75. Yang, Y. H., S. Dudoit, P. Luu, D. M. Lin, V. Peng, J. Ngai, and T. P. Speed.2002. Normalization for cDNA microarray data: a robust composite methodaddressing single and multiple slide systematic variation. Nucleic Acids Res.30:e15.

76. Ye, R. W., W. Tao, L. Bedzyk, T. Young, M. Chen, and L. Li. 2000. Globalgene expression profiles of Bacillus subtilis grown under anaerobic condi-tions. J. Bacteriol. 182:4458–4465.

77. York, G. K., and R. H. Vaughn. 1964. Mechanisms in the inhibition ofmicroorganisms by sorbic acid. J. Bacteriol. 88:411–417.

78. Zakrzewska, A., A. Boorsma, S. Brul, K. J. Hellingwerf, and F. M. Klis. 2005.Transcriptional response of Saccharomyces cerevisiae to the plasma mem-brane-perturbing compound chitosan. Eukaryot. Cell 4:703–715.

79. Zhang, S., and W. G. Haldenwang. 2005. Contributions of ATP, GTP, andredox state to nutritional stress activation of the Bacillus subtilis �B tran-scription factor. J. Bacteriol. 187:7554–7560.

VOL. 190, 2008 B. SUBTILIS SORBIC ACID STRESS RESPONSE 1761