Mycofumigation by the Volatile Organic Compound-Producing Fungus Muscodor albus Induces Bacterial Cell Death through DNA Damage

Abstract

Muscodor albus belongs to a genus of endophytic fungi that inhibit and kill other fungi, bacteria, and insects through production of a complex mixture of volatile organic compounds (VOCs). This process of mycofumigation has found commercial application for control of human and plant pathogens, but the mechanism of the VOC toxicity is unknown. Here, the mode of action of these volatiles was investigated through a series of genetic screens and biochemical assays. A single-gene knockout screen revealed high sensitivity for Escherichia coli lacking enzymes in the pathways of DNA repair, DNA metabolic process, and response to stress when exposed to the VOCs of M. albus. Furthermore, the sensitivity of knockouts involved in the repair of specific DNA alkyl adducts suggests that the VOCs may induce alkylation. Evidence of DNA damage suggests that these adducts lead to breaks during DNA replication or transcription if not properly repaired. Additional cytotoxicity profiling indicated that during VOC exposure, E. coli became filamentous and demonstrated an increase in cellular membrane fluidity. The volatile nature of the toxic compounds produced by M. albus and their broad range of inhibition make this fungus an attractive biological agent. Understanding the antimicrobial effects and the VOC mode of action will inform the utility and safety of potential mycofumigation applications for M. albus.

INTRODUCTION

The endophytic fungus Muscodor albus (CZ-620) inhibits growth of a broad range of pathogenic fungi and bacteria, as well as some nematode and arthropod species (1–6). The inhibition is achieved exclusively through a complex mixture of volatile organic compounds (VOCs) that M. albus secretes into the headspace of the culture. The volatile compounds emitted by M. albus and other closely related organisms in the genus consist of a combination of short-chain alcohols, organic acids, esters, ketones, and several aromatic hydrocarbons as monitored by gas chromatography-mass spectrometry (GC-MS) (6). The compounds range from two to nine carbons and include both straight and branched-chain varieties. The larger aromatic products are predicted to be sesquiterpenes and derivatives of naphthalene and azulene but have not been confirmed by comparison to standards. Although many fungal species have been reported to produce VOCs, none have demonstrated the wide-ranging bioactivity seen with isolates of M. albus (7, 8).

The biological function of the toxic compound production by M. albus is unknown. M. albus was first identified as an endophyte, an organism that lives within the inner tissues of plants (9). Although endophytic interactions between fungi and host plants are typically asymptomatic for part, or all, of their life cycle, these can also be commensal relationships. There are examples where fungi improve the host plant's growth, fitness, or response to stress response (10–12). In the case of M. albus, its VOC emissions may serve as a defense mechanism for the host plant against insects or potential pathogens, and these same products have been hypothesized to help M. albus survive by preventing colonization of the host plant by organisms competing for the same environmental niche (6).

Regardless of the biological purpose, the toxicity of M. albus to other organisms is being harnessed for commercial uses. “Mycofumigation” is a process wherein M. albus VOC production inhibits pathogen growth in agricultural seeds, plants, and soil (3, 5, 13). Mycofumigation has also been applied to fruit storage and transportation, where the presence of M. albus increases shelf life and alleviates pressure for expedited shipping (2). Agraquest, a company recently acquired by Bayer, is exploring the use of M. albus in numerous agricultural applications, including, but not limited to, control of fungal and bacterial pathogens in postharvest and soil diseases, building mold remediation, and seed/grain sanitation (U.S. patent application number 20,120,058,058). In particular, the company seeks to use M. albus for soil sterilization, replacing the use of methyl bromide, a pesticide that is highly toxic and detrimental to the ozone layer (U.S. Environmental Protection Agency registration number for M. albus, 69592-RL/T/I). The volatile nature of the toxic compounds produced by M. albus and their broad range of inhibition make this fungus an attractive biological agent for these diverse applications.

The emergent commercial use of this organism necessitates a more complete understanding of the underlying biology of M. albus. It appears that the toxicity of M. albus exposure results from the combined action of more than one of the secreted compounds. None of the individual compounds or classes of compounds alone mimics the toxicity of M. albus (6). However, the individual compounds produced by M. albus do have antimicrobial effects at higher levels. Each class of compound produced by M. albus was evaluated for toxicity using artificial mixtures. On a comparative basis, the esters had more inhibitory activity than any other group (6). Alcohols, including ethanol and larger alkanols, are toxic due to their role in membrane disruption, which leads to the dissipation of the proton gradient (14, 15). The unidentified, higher-molecular-weight VOCs are predicted to be polycyclic sesquiterpenes (molecular formula C₁₅H₂₄), naphthalene derivatives, or azulene derivatives. Naphthalene has long been used as a household fumigant. The compound is toxic at high concentrations to both the adult and larval forms of many moths, though bacteria have no growth inhibition at concentrations as high as 50 mg/liter (16). Azulene is slightly more toxic to the Gram-negative bacterium Vibrio fischeri, with a 50% inhibitory concentration (IC₅₀) value of 1.5 mg/liter (17). It has been proposed that these compounds act synergistically in the context of the mixture produced by the fungus (6). It may be due to this combination of toxic agents that M. albus is lethal to such a broad range of organisms.

Here, we explore the basis of M. albus VOC toxicity using genetic screens to identify the pathways affected by the VOCs. This is coupled with cytotoxicity profiling to monitor bacterial response to VOC exposure. It appears that the VOC mixture targets diverse cellular pathways. These results explain the modes of action and complement the known broad toxicity of the fungus. This analysis will inform the use of M. albus VOCs for mycofumigation.

MATERIALS AND METHODS

The fungal culture Muscodor albus CZ-620 was obtained through Gary Strobel at Montana State University. All fungal cultures were maintained on potato dextrose agar (PDA) plates (24 g potato dextrose broth [EMD], 15 g agar [BD Difco], 1 liter distilled water) stored at 23°C. Cultures were propagated by transfer of 5-mm culture plugs derived from these plates.

Strains of Escherichia coli were obtained from the American Type Culture Collection (ATCC) (Manassas, VA), ATCC 25922 and ATCC 35218. Laboratory and clinical isolates of Acinetobacter, Citrobacter, Enterobacter, Klebsiella, Pseudomonas, and Staphylococcus were generously provided by Thomas Murray (Yale University).

The library of Escherichia coli K-12 knockout clones in the Keio collection (18) was purchased from Thermo Scientific. E. coli BW25113 was acquired from the Yale Coli Genetic Stock Center (Yale University) and used as the control wild-type E. coli strain.

Bacterial cultures were maintained on autoclaved Luria broth (LB) (10 g tryptone, 5 g yeast extract, 10 g NaCl, 15 g agar, 1 liter distilled water) plates grown at 37°C and stored at 4°C. When necessary, LB was supplemented with 100 μg/ml ampicillin or 25 μg/ml kanamycin. For longer storage, 750 μl of overnight culture grown in LB was mixed with 250 μl 80% glycerol, flash frozen in liquid nitrogen, and stored at −80°C.

Restriction enzymes and DNA ligase were purchased from New England BioLabs (Ipswich, MA, USA). Plasmid pHC79 was obtained from the ATCC (Manassas, VA). Plasmid pUC19 was obtained from New England BioLabs. Ampicillin and kanamycin were obtained through Sigma Chemical Company (St. Louis, MO). Sytox Green stain is commercially available through Life Technologies.

Growth inhibition assays. (i) CFU survival.

The survival of each bacterial variant was measured after exposure to M. albus VOCs by assaying CFU (19). M. albus was inoculated on a 10-ml PDA slant in a 20-ml glass GC-MS vial fused to an identical 20-ml glass GC-MS vial (split vial system) and grown at 23°C for 3 days. Overnight cultures of test bacteria, grown with constant agitation of 150 rpm at 23°C, were diluted in fresh LB medium to an optical density (OD) of 0.05. Diluted cultures were exposed to 3-day-old M. albus headspace within the split glass vial system. Diluted cultures were placed within the split glass vial system, opposite the 10-ml slant of PDA with day 3 M. albus. This split vial was maintained at a constant agitation of 150 rpm at 23°C with no physical contact between the two cultures, only open headspace. One-hundred-fifty-microliter aliquots of bacterial culture were taken at intervals, and serial dilutions were plated onto LB agar plates. Samples were removed by syringe through a rubber septum so as to minimize VOC dispersal from the system. The CFU of the postexposure bacterial culture were quantified after growth at 37°C overnight. The ratio of CFU after exposure to M. albus VOCs to the CFU when M. albus was absent was used to determine percent “survival.”

For CFU survival of E. coli 25922 in stationary phase, the same process as described above was executed with the following exceptions. Overnight cultures of bacteria were not diluted but rather directly transferred to the split vial system for VOC exposure.

(ii) Kinetic growth profile.

The growth of bacterial variants was measured using a BioTek Synergy 4 hybrid microplate reader (Winooski, VT). Five-millimeter plugs of M. albus were inoculated on 5-ml PDA slants in the surrounding wells of a 12-well plate (Costar 3513; Corning, NY). Plates were incubated at 23°C for 3 days, after which bacteria were inoculated in the internal wells and exposed to the VOCs. Overnight cultures of bacteria, grown at 23°C, were diluted to an OD of 0.05, and 2 ml of the diluted culture was added to the internal wells. Plates were sealed using Petri-Seal adhesive sealing film (Sigma-Aldrich) and incubated with a constant agitation of 150 rpm at 23°C. Their growth was recorded spectrophotometrically at 600 nm every 3 min for the incubation time of 30 h. Bacterial growths with noninoculated PDA slants in the surrounding wells were used as controls in this study.

Genomic library screen: isolation and digestion of genomic DNA (gDNA) from E. coli ATCC 35218.

E. coli agarose plugs were prepared as described by Bio-Rad (20) with modifications. Overnight cultures grown at 23°C were diluted to an OD of 0.05 and treated with M. albus VOCs at a constant agitation of 150 rpm for 8 h. At the designated time, 5 × 10⁸ cells/ml were washed at 4°C with resuspension buffer (10 mM Tris-HCl, pH 7.2, 20 mM NaCl, 50 mM EDTA, pH 8.0) and resuspended to half the volume of plugs desired. Cultures were equilibrated to 50°C. A 1× Tris-borate-EDTA (TBE)–2% low-melting-point (LMP) agarose mixture (Lonza, Allendale, NJ) was solubilized by heating and equilibrated to 50°C (final plug solution was 1%). After equilibration of both solutions to 50°C, equal amounts of 2% LMP agarose and cell suspension were mixed by gentle agitation. One hundred microliters was pipetted into each plug mold and set at 23°C for 30 min to solidify. Plugs were then incubated at 37°C overnight in lysozyme buffer (10 mM Tris-HCl, pH 7.2, 50 mM NaCl, 0.2% sodium deoxycholate, 0.5% sodium lauryl sarcosine). Overnight buffer was then removed, and samples were placed in proteinase K solution (100 mM EDTA, pH 8.0, 0.2% sodium deoxycholate, 1.0% sodium lauryl sarcosine, 1 mg/ml proteinase K) at 50°C for at least 4 h or overnight.

Genomic DNA was digested while in the plugs to minimize shearing of DNA and confirmed by pulsed-field gel electrophoresis (PFGE). Plugs were washed with 1× digestion buffer (NEB), and the restriction enzyme PstI or BamHI (30 to 50 U per 100-μl plug) was added for 5 h at 37°C. Following digestion, plugs were washed three times with 50 mM NaCl, and digestion was confirmed by PFGE. GELase buffer (Epicentre Biotechnologies) was added (2 μl per 100-μl plug), and the solution was melted at 68°C.

After cooling to 45°C, GELase enzyme was added and mixed by gentle agitation (1 U per 300-μl melted solution) (Epicentre Biotechnologies). This mixture was incubated at 45°C for over 1 h. Samples were then extracted using phenol-chloroform, and the digested DNA solution was precipitated in ethanol. Digested DNA was inserted into the PstI or BamHI site of vector pHC79 (ATCC 37030).

Packaging of E. coli 35218 library for amplification and screening within ElectroMAX DH10B.

Packing of the E. coli 35218 library in pHC79 was performed as described in the instruction manual for Gigapack III Gold packaging extract (Agilent Technologies) with few modifications. Cosmid pHC79 and E. coli DNA library were packaged in phage as described above, and a 1:50 dilution was mixed with prepared host strain E. coli VCS257. The titer of the packaging reaction mixture was determined to ensure 20× coverage of library and plated on LB agar plates supplemented with ampicillin, and viable colonies was harvested. The library was isolated using the low-copy-number cosmid protocol of the Qiagen maxikit (Qiagen, Limburg, The Netherlands).

Electroporation of the library into ElectroMAX DH10B cells (Invitrogen) was performed according to Invitrogen instructions. Transformants were grown overnight at 37°C at serial dilutions to maintain single colonies. Overnight plates were then replica plated onto LB agar plates, and lids were removed and inverted onto 3-day-grown M. albus. Inverted plates were sealed and incubated at 23°C overnight. Viable colonies were streaked upon LB agar plates supplemented with ampicillin and grown overnight at 37°C.

Cosmids from viable transformants were isolated using the Qiagen maxikit with the low-copy-number cosmid isolation protocol. Primer walking was performed, and genomic sequence was identified (Genewiz). Identified genes were isolated using PCR from the genomic fragment and inserted into pUC19. Individual genes were transformed in ElectroMAX DH10B cells and screened in the same manner as the library.

Independent expression of RecA in E. coli DH10B.

Using primers pUC19RecaFW and pUC19RecaRV, the recA gene was cloned from E. coli 35218 genomic DNA and inserted into high-copy-number plasmid pUC19. ElectroMAX DH10B cells were electroporated with the pUC19 recA construct as described above. Transformants were then grown against M. albus as described in the plate reader assay above.

Knockout library screen.

Clones from the Keio collection were provided in 96-well microtiter plates and stored at −80°C. For propagation, all knockout strains were grown in LB containing 25 μg/ml kanamycin at 37°C. When grown against fungal samples, knockout strains were grown at 23°C on LB agar containing kanamycin as described above.

Screening was performed using a 5- by 5-mm² culture plug of M. albus inoculated on a one-well PDA plate of a size identical to the 96-well microtiter plate (Costar 3593 [Corning, NY]). The fungal sample was grown at 23°C for 3 days prior to bacterial exposure.

For screening of knockout strains, each 96-well microtiter plate was temporarily removed from the −80°C storage. Without thawing, a 96-pin replicator was used to inoculate a 96-well microtiter plate (Costar 3593 [Corning, NY]) containing 200 μl LB and kanamycin per well. Stock plates immediately returned to −80°C. Inoculated plates were then put at 37°C for static growth overnight. At the same time as plate inoculation, cultures of JW1314 (ΔrecT) and JW2669 (ΔrecA) were each inoculated in 5 ml LB with kanamycin, for controls. These samples were grown at 37°C with constant agitation overnight.

The 96-well plate overnight culture was then pinned on LB agar supplemented with kanamycin. In addition, 2 μl of cultures JW1314 (ΔrecT, known VOC-resistant knockout) and JW2669 (ΔrecA, known VOC-sensitive knockout) were spotted on the same plate to serve as controls. Lids from both this inoculated LB-kanamycin plate and the 3-day-old culture of M. albus were removed, and cultured plates were inverted over each other and sealed using Petri-Seal adhesive sealing film (Sigma-Aldrich). Cultures were grown overnight at 23°C.

Cultures were analyzed for growth after 24 h of exposure to M. albus by visual observation for degree of culture spot growth. Each plate was screened against M. albus three times to ensure reproducibility of the results. Plates in which JW2669 (ΔrecA) was not inhibited by M. albus or in which JW1314 (ΔrecT) was inhibited were excluded from the analysis.

GO analysis.

We used ontologies from Gene Ontology (GO) (21) (http://www.geneontology.org/ontology/gene_ontology.obo, version 2014-04-30), while annotations were obtained from EcoCyc for Escherichia coli strain MG1655 (22). Analysis was performed using a custom-built algorithm in MATLAB, including a hypergeometric test to compute P values which were subsequently adjusted with the Benjamini-Hochberg-Yekutieli false discovery rate procedure (23, 24). MATLAB codes are available upon request.

Microscopy.

E. coli cells were grown for more than 5 generations, up to exponential phase (OD at 600 nm [OD₆₀₀] of ∼0.1) at 25°C with constant agitation. The cultures were then exposed to the fungus M. albus for the specified amount of time and spotted on agarose pads before imaging at room temperature with an inverted Eclipse80i microscope (Nikon, Tokyo, Japan) equipped with a Hamamatsu Orca-ER camera and phase-contrast objective Plan Apochromat 100× and a 1.40 numerical aperture (NA). Images were acquired using the MetaMorph software from Molecular Devices (Sunnyvale, CA, USA). DAPI (4′,6-diamidino-2-phenylindole) labeling was performed 30 min prior to imaging (1 μg/ml). Cellular length and DAPI intensity were measured using MicrobeTracker (25).

NPN-uptake permeability assay.

Cell permeability using the dye 1-N-phenylnaphthylamine (NPN) (Sigma-Aldrich) was measured as previously described (26). E. coli was grown for more than 5 generations, up to exponential phase (OD₆₀₀ of ∼0.1), in LB with an appropriate antibiotic at 23°C. Cultures were then treated under the various conditions and grown for another 3 h, with shaking, at 23°C. After 3 h of exposure, samples were washed and diluted to an OD of ∼0.7/0.8. An aliquot of cells at this point was removed and plated for CFU measurement. Polymyxin B was added at 1/2 MIC (0.5 μg/ml) and allowed to incubate for 5 min before measurement. Cell suspensions were analyzed using a PTI-814 fluorescent spectrophotometer (PTI, New Jersey). Fluorescence was detected with an excitation of 350 nm and an emission of 420 nm with split widths of 4 nm. Fluorescence was then normalized by the CFU data for each condition, giving relative fluorescence (relative fluorescent units [RFU]) of each sample.

Pulsed-field gel electrophoresis for DNA damage detection.

Agarose plugs for pulsed-field gel electrophoresis were prepared as described above under “Genomic library screen: isolation and digestion of genomic DNA (gDNA) from E. coli ATCC 35218.” After incubation with proteinase K solution, samples were incubated in 1× TBE to prepare for electrophoresis. Samples were run on a 1% agarose (pulsed-field-certified agarose; Bio-Rad) gel for 7 h at 14°C, 6 V, initially to a final switch of 1 to 6 V. The gel was stained using a 10,000-fold dilution of SYBR Gold nucleic acid stain (Life Technologies) for 40 min and imaged using Chemidoc MP (Bio-Rad).

Gel electrophoresis for analysis of DNA nicks and strand breaks.

Gel electrophoresis was performed on plasmid pBR322 (New England BioLabs Inc.) as previously described (27). Supercoiled pBR322 was independently treated with Nt.BsmI nicking enzyme and restriction enzyme BamHI (New England BioLabs Inc.). Each sample (0.5 μg) was loaded on a 1.2% agarose gel and run at 6 V/cm. The gel was then stained using SYBR Gold (Life Technologies) at a 1:10,000 dilution.

RESULTS

Identifying sensitive and resistant bacterial strains.

We set out to identify the modes of action (MOA) of M. albus VOC toxicity by screening a collection of Escherichia coli isolates (Thomas Murray [Yale University] and ATCC) for viability after treatment with M. albus VOCs. The collection consisted of seven E. coli strains, derived from both E. coli B and E. coli K-12, including clinical lab isolates and cultured isolates (Table 1). Using the simple bioassay system devised to examine only volatile agents for microbial inhibition (6), test E. coli cultures were monitored for growth during exposure to the M. albus VOCs. No visible growth during exposure was observed for five of the seven strains tested. After 24 h, the VOCs were removed, and the absence of further growth revealed that the inhibition was bactericidal to these five strains (Table 1). Of these sensitive strains, we selected the E. coli strains ATCC 25922 and DH10B for further analysis. E. coli 25922 is a well-characterized strain that is a control Gram-negative bacterium widely used for various laboratory experiments, especially for antibiotic susceptibility assays (28). The E. coli strain DH10B was selected because it is efficient for DNA cloning. Two of the test bacteria, E. coli ATCC 35218 and E. coli K-12 strain BW25113, remained viable after 24 h of VOC exposure despite an initially static growth inhibition. E. coli ATCC 35218 is also a commonly used ATCC strain for antimicrobial susceptibility studies, and BW25113 is the parent strain for the Keio collection of single-gene knockouts (18).

TABLE 1.

Bacterial strains used in this study

E. coli strain name	Genotype	Source	Susceptibility to M. albus VOCs
HB101	F− thi-1 hsdS20 (rB− mB−) supE44 recA13 ara-14 leuB6 proA2 lacY1 galK2 rps L20 (Strr) xyl-5 mtl-1	CGSC (Yale University)	Sensitive
ElectroMAX DH10B	F− endA1 recA1 galE15 galK16 nupG rpsL ΔlacX74 ϕ80lacZΔM15 araD139 Δ(ara, leu)7697 mcrA Δ(mrr-hsdRMS-mcrBC) λ−	Life Technologies	Sensitive
Rosetta 2	F− ompT hsdSB(rB− mB−) gal dcm λ(DE3 [lacI lacUV5-T7 gene 1 ind1 sam7 nin5]) pLysSRARE (Camr)	EMD Millipore Merck	Sensitive
25922	FDA strain Seattle 1946 (DSM 1103, NCIB 12210)	ATCC	Sensitive
Clinical isolate	Unknown	Murray lab (Yale University)	Sensitive
K-12 BW25113	rrnB3 ΔlacZ4787 hsdR514 Δ(araBAD)567 Δ(rhaBAD)568 rph-1	CGSC (Yale University)	Tolerant
35218	Strain designation 1532	ATCC	Tolerant

Quantitation of M. albus VOC inhibition was accomplished by measuring CFU for each strain during exposure to the VOCs. E. coli 25922 was completely susceptible to the VOC exposure and steadily lost viability during treatment (Fig. 1). Tolerant strain E. coli BW25113 exhibited static inhibition of growth upon VOC exposure; however, following the initial hours of exposure, E. coli BW25113 resumed exponential growth. Similarly to BW25113, E. coli 35218 was viable after 24 h of M. albus VOC exposure. In 35218, however, the initial growth inhibition was much more severe as the culture viability steadily decreased over the first few hours of VOC treatment, and then growth recovered and mimicked that of untreated cultures (see Fig. S1 in the supplemental material).

FIG 1.

Growth and inhibition of E. coli upon treatment with M. albus VOCs. Growth of K-12 E. coli BW25113 (top) (blue) and of E. coli 25922 (bottom) (blue) as monitored by CFU assay and after exposure to the M. albus VOCs (red).

Identifying modes of action.

The dramatic differences in growth among the different E. coli strains suggested that a genetic variation between the strains was responsible. We set out to use these differences in VOC susceptibility for a series of genetic screens to identify the genes that play a role in E. coli resistance.

We first determined if a specific genetic element was sufficient to render a bacterial strain tolerant to M. albus VOCs. We constructed a plasmid-based library of the genomic DNA from the tolerant E. coli strain 35218 and screened for resistance in the highly sensitive E. coli strain DH10B. We cloned fragments of E. coli 35218's genome in DH10B and screened the transformants for viability after exposure to M. albus VOCs. The screen resulted in four viable clones out of the approximately 3,000 tested. We extracted the library plasmid from each of the four viable clones and sequenced the genomic library fragments to identify the individual genes contained in each fragment. The gene encoding the DNA repair protein recombinase A (RecA) was present in all four of the isolated clones. An independent construct containing solely the recA gene was sufficient to rescue viability of E. coli DH10B after VOC exposure for 24 h. This transformant initially demonstrated static growth during exposure to VOCs; however, after a single doubling time it resumed exponential growth (Fig. 2). No other gene isolated from the clones was sufficient to induce resistance in DH10B. Isolation of a single repair gene demonstrated that DNA repair is an essential process for viability against M. albus. In the genetic background of DH10B, RecA is sufficient for viability. DH10B is a highly efficient competent cell line that contains the recA1 allele, which has a single point mutation that renders the strain defective in all known in vivo functions of the recA gene. This increases its utility as a cloning strain as it reduces the occurrence of unwanted recombination of cloned DNA (29). However, the highly sensitive strain E. coli 25922, in addition to other clinical and lab isolates (Table 1), contains functioning recA genes and functional DNA repair pathways and yet is still sensitive to M. albus VOCs. This result led us to suspect that while DNA repair is required for resistance, it is not the only component necessary for tolerance to the VOC mixture.

FIG 2.

Growth curve of E. coli DH10B and E. coli DH10B expressing recA during treatment with VOCs. (a) DH10B growth curve (black) as measured spectrophotometrically by OD₆₀₀ and upon treatment with M. albus VOCs (red). (b) Growth curve of DH10B expressing pUC19 recA (black) and during treatment with M. albus VOCs (red). Data are representative of at least three biological replicates.

Given that a single gene could confer resistance to the M. albus VOCs, we sought to identify all the genes that contribute to tolerance in E. coli. We screened the Keio collection of single-gene-knockout strains in the background of the tolerant E. coli strain BW25113. The inhibitory effect of VOC exposure on the single-gene knockouts was scored as follows: complete inhibition, slight inhibition, or no effect relative to the unexposed controls. Out of the 3,869 known-open-reading-frame (ORF) knockouts, a total of 141 gene knockouts had increased sensitivity to VOC exposure. Twenty-seven of the single-gene knockouts were completely inhibited by M. albus VOCs. All the knockouts that demonstrated increased sensitivity to M. albus VOCs were categorized using Gene Ontology (GO) (21). Using these GO classifications, we performed a gene-set enrichment analysis. This classified the mutants into a total of 17 GO categories that were enriched among knockouts sensitive to M. albus VOCs (Fig. 3). The most enriched categories were DNA repair, DNA metabolic process, and response to stress. Additional categories, such as response to DNA damage, SOS response, macromolecule metabolic processes, and DNA recombination, were also prominent in the sensitive set. The enrichment analysis supported our initial observations that DNA repair was necessary for VOC tolerance, but it also highlighted other pathways not previously considered.

FIG 3.

Gene Ontology category enrichment analysis. Gene Ontology (GO) category classification of gene knockouts demonstrating increased sensitivity to treatment with M. albus VOCs. Enrichment analysis calculated using a 1% cutoff. GO categories are listed on the y axis in descending order of enrichment. Negative log values of the P value for category enrichment are displayed on the x axis.

Sensitive knockouts include the DNA repair genes recA, -B, -C, -D, -G, -F, -J, -N, -O, and -R and 16 additional genes involved in DNA repair, including uvrD and ruvA. Approximately 40% of the genes involved in DNA repair increased the sensitivity to M. albus VOCs when absent. Among these were genes involved in homologous recombination (HR), nucleotide excision repair (NER), base excision repair (BER), and mismatch repair (MMR), each of the DNA repair categories in bacteria. The second most enriched category involved genes that respond to cellular stress, including those for many envelope stress proteins, outer membrane transporters, proteins that respond to oxidative reagents, and proteins involved in the response to extracellular stimuli, including ethanol and osmotic shock. The knockouts of SOS response genes, including ruvA, ruvV, dinB, recA, recN, symE, recF, uvrA, uvrD, and uvrC, all demonstrated increased sensitivity to VOCs. Eleven of the 141 sensitive knockouts were involved in protein and RNA degradation, including the Lon and other proteases, ClpX, ElaD, HyaD, and GlgG. Unexpectedly, the gene knockouts of four cell division proteins (MinC, Ttk, YcbG, and ZapA) and six proteins involved in chemotaxis (CheY, FlgC, FliJ, Flk, YraI, and YcbU) also demonstrated increased sensitivity to the M. albus VOCs. Our results seem to indicate a large role for cellular stress responses and DNA repair in M. albus tolerance; however, the diversity of other genes suggests a complex MOA with multiple targeted pathways.

M. albus VOCs induce DNA damage.

We sought to verify the direct role of M. albus VOCs in these target pathways. We monitored tolerant and sensitive bacterial strains exposed to M. albus VOCs for phenotypic changes in cellular morphology, membrane fitness, and DNA integrity.

Live-cell imaging showed that after a 3-hour exposure, the VOCs produced by the fungus resulted in filamentation of the sensitive E. coli 25922 cells (Fig. 4A and B) and, to a lesser extent, of the tolerant E. coli 35218 cells (see Fig. S2A and B in the supplemental material). With longer treatment, the morphology of the tolerant E. coli 35218 strain was similar to that of nonexposed cells (data not shown). A filamentous phenotype in E. coli could be due to a direct effect on the division machinery or to an indirect inhibition of cell division. When cell division is directly inhibited with the β-lactam antibiotic cephalexin, DNA replication and segregation occur normally, resulting in multiple nucleoids evenly distributed along the cell filament (data not shown and reference 30). In contrast, after DAPI staining, we observed that the DNA of the sensitive cells exposed to M. albus remained confined, forming one centrally located nucleoid flanked by large DNA-free regions (Fig. 4A; see also Fig. S2A in the supplemental material). In addition, the DNA amount per cell, as measured by the DAPI signal intensity, was barely higher in exposed cells than in VOC-untreated cells, despite a large difference in cell size (Fig. 4B and C). As a consequence, the DNA concentration per cell was considerably smaller in cells exposed to VOCs (Fig. 4D). This difference was much less noticeable for the tolerant strain (see Fig. S2C and D in the supplemental material). These results are consistent with the idea that VOCs induce DNA damage and trigger related stress responses that result in inhibition of DNA replication and cell division.

FIG 4.

M. albus VOCs affect cellular morphology and DNA integrity in E. coli 25922 cells. (A) Phase-contrast and fluorescence images of E. coli 25922 cells after a 3-hour period at 25°C in the presence (+) or absence (−) of M. albus VOCs. Cultures were DAPI stained (2 mg/ml) for 20 min prior to imaging on 1% agarose pads. Bar, 2 mm. (B) Density probability distribution of cell length in the absence (red line) or presence (orange line) of M. albus. The cell length distribution of E. coli K-12 is shown in gray for comparison. (C) Density probability distribution of the DAPI fluorescence signal per cell in the absence (red) or presence (orange) of M. albus in arbitrary units (au). (D) Density probability distribution of the DAPI mean intensity per cell in the absence (red) or presence (orange) of M. albus in arbitrary units (au).

Antimicrobial agents whose MOA is the inhibition of DNA replication induce death through the accumulation of DNA breaks that either are too abundant to be repaired or cannot be repaired due to inactivated replication or repair machinery. We monitored the formation of DNA breaks using pulsed-field gel electrophoresis (PFGE), where the distribution of randomly broken DNA molecules can be determined by DNA migration from the high-molecular-weight chromosomal DNA band (31, 32). Exponentially growing cultures were exposed to M. albus VOCs and then embedded in a low-melting-point agarose gel in order to prevent the indirect breakage of genomic DNA during the subsequent cell lysis. After analysis by PFGE, the genomic DNA from healthy nonexposed cells was observed as a single, clear band with negligible evidence of DNA fragmentation (Fig. 5a, lanes 2 to 4). In contrast, the genomic DNA isolated from sensitive, exposed strains had significant accumulation of lower-molecular-weight DNA fragments, indicative of double-stranded DNA damage (Fig. 5a, lanes 5 to 7). In sensitive strains, the accumulation of fragments increased with increasing VOC exposure time. This accumulation was not visible in tolerant strain E. coli BW25113 (Fig. 5b), where exposed cells showed negligible evidence of DNA damage, similar to healthy nonexposed cells. This visible accumulation of short DNA fragments in the sensitive strain exposed to VOCs confirmed that double-stranded DNA damage is an important element of M. albus toxicity.

FIG 5.

Pulsed-field gel electrophoresis (PFGE) of genomic DNA from E. coli treated with M. albus VOCs. (a) E. coli 25922 without exposure to M. albus VOCs (lanes 2 to 4) or exposed to M. albus VOCs (lanes 5 to 7) for 3, 4, or 5 h (from left to right). (b) E. coli BW25113 was treated as described above, without VOC exposure (−) (lanes 9 to 11) or with VOC exposure (+) (lanes 12 to 14). The number of cells/plug was normalized by the weight of the cell pellet. Lanes 1 and 8 show 24-kb Max DNA ladders (Fisher).

There are at least two mechanisms to explain this accumulation of DNA damage. The first is an indirect model in which M. albus VOCs interact with cellular compounds or machinery that in turn causes damage to DNA. The second involves direct damage to the DNA in the form of nicks or breaks by M. albus VOCs.

To test if the VOCs are acting by a direct or an indirect mechanism, we exposed purified supercoiled plasmid pBR322 to the M. albus VOCs. The exposed DNA was then analyzed using gel electrophoresis (Fig. 6). Mobility of a supercoiled plasmid through high-concentration agarose does not reflect its linear size due to its tightly wound conformation. A single nick in the plasmid relieves the superhelical tension and causes the DNA to migrate slower in the gel. When completely linearized, the plasmid runs according to its true length relative to a double-stranded DNA size ladder. No visible nicking or cleavage was observed upon VOC treatment of purified plasmid after 24 and 48 h. These data suggest that the DNA damage within sensitive organisms is not due to direct lesions caused by M. albus VOCs. Instead, the damage results from an indirect process.

FIG 6.

Gel electrophoresis analysis for single- and double-stranded DNA breaks of plasmid DNA after exposure to M. albus volatile compounds. Plasmid pBR322 (lanes 1 and 6) was exposed to M. albus VOCs in Tris-EDTA, pH 8.0, for 24 h (lane 2) and 48 h (lane 3). pBR322 was cut with restriction enzyme BamHI (lane 4) and nicking enzyme Nt.BsmI (lane 5) for comparison.

These data, in addition to the observation of DNA replication arrest, suggest that the M. albus VOCs have a toxic effect on bacteria that are actively growing, replicating, and dividing. To test this, we exposed a culture of E. coli 25922 in stationary phase to the VOCs and monitored viable cell counts over time. After 7 h of exposure, approximately 95% of the stationary-phase culture was killed, compared to more than 99.99% of cells in an exponentially growing culture (Fig. 7). The metabolically reduced culture was at least 1,000-fold less sensitive to the VOCs than the exponentially active culture. These data suggest that M. albus VOCs are acting in a manner that is dependent on the metabolic activity of the bacteria.

FIG 7.

Assessment of VOC toxicity effect on E. coli metabolic state. Percent viability of E. coli 25922 cultures exposed to M. albus VOCs for 0 h (black) and 7 h (gray) as measured by CFU. Percent viability was calculated by number of CFU in culture exposed to VOCs compared to the CFU of control culture with no VOC treatment.

M. albus VOCs increase membrane permeability.

In addition to DNA processes, cell envelope fitness is another common target for antibiotics, including permeation of the cell membrane by alcohols, such as those found in the M. albus VOC mixture. Of the 27 gene knockouts belonging to the cellular response to stress category, 15 were specifically related to membrane functions. To test for effects on membrane integrity, we probed for damage of the bacterial outer membrane after exposure to M. albus VOCs using the cell-impermeant dye 1-N-phenylnaphthylamine (NPN). The intensity of the fluorescent probe is correlated with the degree of outer membrane damage and intercalation of the dye into the membrane. We found that untreated healthy cells display a background level of fluorescence due to nonspecific surface binding of the dye (Fig. 8). The fluorescence increased 3-fold upon treatment with polymyxin B (at 1/2 MIC), an antibiotic known to permeate outer membranes of Gram-negative bacteria. Tolerant strain E. coli BW25113 displayed a greater-than-10-fold increase in NPN fluorescence upon exposure to M. albus VOCs for 3 h. Similarly, the sensitive recA knockout of BW25113 demonstrated an increase in permeability upon VOC exposure. These results indicate that M. albus VOCs cause an increase in membrane permeability, independent of strain resistance or ability to repair DNA.

FIG 8.

NPN-uptake permeability of the outer membrane of E. coli upon M. albus VOC exposure. Permeability of E. coli K-12 strain BW25113 (black) and E. coli K-12 strain BW25113 ΔrecA (gray) in response to treatment with Muscodor VOCs. Antibiotics were administered at concentrations corresponding to 1/2 MICs. Permeability was monitored by fluorescence of 1-N-phenylnaphthylamine (NPN) at 420 nm. Relative fluorescent unit (RFU) = intensity of fluorescence/cell density (CFU).

DISCUSSION

M. albus displays multitarget toxicity.

M. albus produces a mixture of VOCs that are lethal not only to a broad range of bacteria and fungi but also to some nematode and arthropod species. This antimicrobial activity stems from one or a combination of the VOCs produced by the fungus, which include alcohols, acids, esters, ketones, and aromatic hydrocarbons (6). The modes of action of these VOCs and the pathways that they target to inhibit bacterial growth were unknown and have been challenging to characterize due to the volatility and complex mixture of the compounds.

How does M. albus inhibit such a diverse group of organisms? Previous studies suggested that the toxicity of M. albus is due to the synergistic activities of multiple antimicrobial compounds (6). Using a combination of genomic knockout viability, gain-of-function screening, and comparative growth inhibition assays, we have demonstrated that M. albus toxicity is the result of VOCs permeabilizing the cell membrane and inducing disruption of cellular DNA metabolism through DNA damage. Our data support the multiple-component MOA hypothesis proposed by initial studies on M. albus (6) and identify the particular pathways targeted by the complex VOC mixtures.

Increase in membrane permeability.

The Gene Ontology category “cellular response to stress” includes outer membrane proteins related to cellular envelope stress and membrane transporters. The outer membrane of Gram-negative bacteria is an efficient barrier to noxious compounds, antibiotics, and hydrophobic compounds. A selective permeation barrier is notably maintained through nonspecific porin channels as well as specific channels such as TonB-dependent siderophore transports (33). Disruption of these channels can lead to increased susceptibility to antimicrobials. In our screen, E. coli BW25113 knockouts of genes encoding nonspecific channels or porins and specific transporters such as TonB and MdtK were strongly inhibited by VOCs.

Membrane stress could also derive from direct interaction of compounds with the lipid bilayer. Compounds such as butanol and ethanol are known cell membrane disrupters as they disorder the physical structure of the membrane (34). M. albus produces short-chain alcohols that could have this activity. These short-chain, branched alcohols are known to have toxic effects on E. coli due to increased permeability when present at high concentrations (15, 35). In our screen, the knockout of 44 genes functioning as or in association with membrane proteins rendered the E. coli strain highly sensitive to M. albus VOCs. The involvement of all these membrane-related proteins emphasizes the importance of maintaining a selective barrier against the VOC mixture and gives insight into part of the M. albus toxic MOA on the bacterial outer membrane.

As previously mentioned, 27 of the sensitive gene knockouts belonged to the “cellular response to stress” category. Six of these knockouts, fumC, hscB, iscA, iscU, ygfX, and sodC, are involved in the cellular stress response to oxidative reagents. Reactive oxygen species (ROS) have recently been implicated in the mechanism of cellular death by bactericidal antibiotics (36, 37). Although the role of ROS in the toxicity of antibiotics has been debated, an increase in the intracellular levels of ROS has been measured upon treatment with bactericidal antibiotics under aerobic conditions (36, 38–42). For example, the antibiotic ampicillin primarily targets the biosynthesis of the peptidoglycan cell wall; however, it also induces increased levels of ROS that correspond to initiation of the SOS response, implicating DNA damage (36). Similarly, n-butanol increases membrane fluidity, while at the same time inducing a large increase in ROS (43). This intracellular increase can lead not only to DNA damage through sugar modifications but also to inactivation of enzymes and oxidation of amino acids and polyunsaturated fatty acids (44–47). In all cases, this increase in oxidative stress is the cellular response to the extracellular xenobiotics and highlights the importance of mitigating the interactions of ROS and cellular DNA (48).

Interruption of DNA metabolism.

The most prevalent consequence of M. albus VOC exposure is disruption of DNA metabolism. This is evident in the phenotypic changes of VOC-exposed E. coli. The elongation of cell length with no increase in the amount of DNA per cell is a well-documented effect of DNA replication inhibition (49–51). In E. coli, cell division and DNA replication are closely coordinated. When DNA replication is interrupted, bacteria initiate the SOS response, which causes an immediate arrest of cellular division and leads to the filamentous phenotype (50). This is a clear and distinct phenotypic effect seen for all cells when exposed to M. albus VOCs. Induction of the SOS response involves derepression of genes involved in DNA repair, DNA mutagenesis, and the inhibition of cell division. The SOS response acts as an inducible DNA repair system to survive high levels of DNA damage. This response plays a major role in VOC tolerance, as the knockouts of 10 SOS response proteins were highly sensitive to the M. albus VOCs. The necessity of DNA repair and SOS proteins for tolerance of M. albus VOCs clearly demonstrates an MOA involving a disruption of DNA metabolism.

Further supporting the role of VOCs in DNA disruption, a single DNA repair protein, RecA, was the only hit from the functional genetic screen of a tolerant strain. This indicates that in DH10B functional DNA repair systems are necessary for survival under M. albus VOCs. Although expression of recA conferred tolerance on a sensitive strain, DH10B, DNA repair is not sufficient to confer resistance in all sensitive strains. For example, E. coli 25922 was completely inhibited by M. albus VOCs, but based on available strain information, all DNA repair mechanisms and SOS proteins (including RecA and Lon) are functional. The DNA repair system that is sufficient for one E. coli strain is not sufficient for all. This conclusion and the enrichment within 17 gene categories support the multiple-component MOA hypothesis and emphasize the large role played by DNA damage.

The necessity for DNA repair leads to the question of what is causing this damage. Our data suggest that the M. albus VOCs do not cause direct single- or double-stranded breaks to isolated DNA in solution. However, many of the genes from our knockout sensitivity screen suggest that the DNA is being modified. For example, the knockout of DNA methyltransferase Ada was highly sensitive to the VOCs. Ada is a multifunction protein that protects E. coli from methylation damage (52–54). It possesses methyltransferase activity and is converted into a potent transcriptional activator, initiating transcription of the methylation resistance genes (ada, alkA, alkB, and aidB) upon methylation (53). Of these four methylation resistance genes, both ada and alkA were sensitive in our screen. AlkA is a 3-methyl-adenine DNA glycosylase (55). AlkA works to remove not only methyl adducts from adenine but also methyl and ethyl derivatives of guanine and thymine (52, 55). Similarly, the enzyme TAG is a 3-methyl-adenine glycosylase that was sensitive in our screen (52). The requirement of these particular genes for VOC resistance strongly suggests the formation of these lesions, which if not properly repaired would form mutations or lead to breaks during replication (56, 57). Their sensitivities despite the presence of redundant enzymatic functions suggest that the VOCs are inducing the formation of alkyl adducts on the DNA.

The repair enzymes that were not hits in the screen reinforce the importance of the sensitive repair-enzyme knockouts and provide further insight into the VOC-induced DNA damage. For instance, the knockout of DNA glycosylase MutM did not create sensitivity. MutM is primarily responsible for removing oxidized bases due to free radical-induced lesions (58, 59). Similarly, the endonucleases Nfo and Nei, which were not required for resistance, are specific for damaged bases due to specific oxidants (60–62). As this insensitivity could be due to crossover and redundancy of function between enzymes, it reduces but does not completely eliminate the likelihood that oxidized base modifications are formed upon Muscodor VOC exposure.

Current microbe control for plants, soil, and fruit and vegetable transport utilizes fumigants such as methyl bromide, chloropicrin, or phosphine, among others (63). These chemicals are all toxic to humans and often harmful to plants, and methyl bromide causes significant ozone depletion (64). M. albus VOCs have been suggested as an alternative to these broad-spectrum chemical fumigants. Without in-depth analysis, this use of M. albus is attractive as the antimicrobials are “naturally” produced by the organism. These natural products, however, are DNA damaging, and their effect on human DNA is yet to be determined. Understanding the mechanisms of action of the VOCs will help establish the appropriate applications of M. albus as an industrial mycofumigant.