The wealth of bacterial volatiles

Microbiologist have recognized for a long time that bacteria emit characteristic scents, e.g., the typical odor of indole from Escherichia coli or butyric acid and acetone from Clostridium acetobutylicum. Many more odors are known, and humans have exploited microbial volatiles as aroma components of cheese, sauerkraut, yoghurt, wine, etc. Moreover, the repugnant smell of rotting organic matter often results from the release of bacterial volatiles. Although the human nose can distinguish many volatiles, there is a given limitation in the detection and proper description of the relevant smell. Recent investigations using gas chromatography (GC) and mass spectrometry (MS) illustrate the splendid capacity of bacteria to produce a wealth of volatile compounds (Kai et al. 2006; Schulz and Dickschat 2007; Bunge et al. 2008). Whereas the odors and aromas released from bacteria during food transformations and fermentations or from building material have been intensively investigated, only little is presently known about the general ability and efficiency of volatile emissions of bacteria (Dainty et al. 1985, 1989; Korpi et al. 1998).

One of the earliest papers that described the production of volatiles by bacteria (dysentery group) demonstrated the release of formic and butyric acid (Zoller and Clark 1921). Stotzky and Schenck (1976) summarized the volatile organic compounds released from microorganisms and showed that fungi produce a wider variety of volatiles than bacteria, although this might have been attributable to the larger number of studies performed with fungi at that time. Stotzky and Schenck (1976) described bacteria such as Pseudomonas spp. and Streptomyces spp. as being ethylene and hydrogen cyanide producers, Clostridium spp. as emitters of dimethyl disulfide, various short chain acids, 2,3-butanediol, isopentanol, and acetoin, and Agrobacterium radiobacter, A. rhizogenes, Bacillus cereus, Enterobacter aerogenes, E. coli, Micrococcus luteus, Nocardia corallina, Proteus vulgaris, Sarcina lutea, and Serratia marcescens as releasers of unidentified volatiles. Furthermore, mixed bacterial cultures and microbes in soil under aerobic and anaerobic conditions have the potential to produce organic volatiles (volatile organic compound, VOC). In recent years, the technology has developed further, and the identification and quantification of volatile compounds has been mostly successful. More than 120 different compounds are emitted from actinomycetes (Schöller et al. 2002), comprising alkanes, alkenes, alcohols, esters, ketones, and isoprenoids. Myxococcus xanthus has also turned out to be a rich source of volatile compounds, 42 compounds having been collected in a closed-loop stripping apparatus (Dickschat et al. 2004). Two new natural products, (S)-9-methyl-decan-3-ol and 9-methyldecan-3-one, have been identified. With an enormous effort and critical evaluation of the published data, Schulz and Dickschat (2007) have summarized all known bacterial compounds so far detected by using the state-of-the-art methodologies, with 346 different compounds released from various bacteria being described. The classification of bacterial volatiles has revealed 75 fatty acid derivatives, 50 aromatic compounds, 74 nitrogen-containing compounds, 30 sulfur compounds, 96 terpenoids, and 18 halogenated, selenium, tellurium, or other metalloid compounds. Several groups of compounds seem to be especially widespread among bacteria, for example, pyrazines, volatile sulfur components, geosmin, and 2-methylisoborneol. Our investigations with bacterial isolates of Stenotrophomonas, Serratia, Pseudomonas, Bacillus, Burkholderia, Erwinia, Agrobacterium, Staphylococcus, and Xanthomonas species have also indicated the emission of complex bacterial blends of odors, comprising in some cases up to 60 compounds per strain (Kai et al. 2006; Kai/Piechulla, unpublished results). Interestingly, the majority of volatiles cannot unequivocally be identified using the NIST-GC-MS library or others, suggesting that for many of them, their structures remain to be elucidated. Most emitted compounds are species-specific, but overlapping volatile patterns have been found for Serratia spp. and Pseudomonas spp., indicating that at least in some cases, these volatile profiles or a typical individual compound (characterized by the retention index until the structure is elucidated) can be used as a biomarker (Kai et al. 2006; Henis et al. 1966). Bacterial volatiles are compiled in a publically available “Super Scent” database (Dunkel et al. 2009).

Volatile detection and identification and methodical constraints

State-of-the-art detection and determination of volatiles are generally performed with the headspace technique and gas chromatography and mass spectrometry (GC-MS). Some limitations of the method should be considered here. GC-MS is a sensitive method that allows the determination of the number of different compounds, their relative quantities in the volatile mixture, and also compound identification. However, compounds can only be considered as identified if they demonstrate identical Kovats indices on two columns of different polarity and display mass spectra coincident with the library. We and others have often observed that the available MS libraries [e.g., National Institute of Science and Technology (NIST), Whileys] do not contain compounds that were emitted by bacteria. Therefore, new compounds have to be structurally elucidated by chemists using, e.g., nuclear magnetic resonance analysis. It also should be kept in mind that headspace collections with absorbance material such as tenax, super Q, or charcoal, etc. favor the binding of compounds with specific chemical features and involve extraction with organic solvents (e.g., dichlormethane, pentane, hexane). The huge bulk of solvent leads to insufficient resolution of highly volatile, early eluting compounds. The use of the solid phase micro-extraction method is a well-accepted alternative, which can also be used to resolve small inorganic or organic volatiles such as CO, CO2, NH3, HCN, and ethylene.

The appearance of a characteristic volatile profile or compound is attributable to the specific metabolism or metabolic pathway(s) that are active in the bacteria. Depending on the growth media and growth conditions, the bouquet of released compounds can vary, e.g., the growth of Stenotrophomonas rhizophila P69 on nutrient broth with and without glucose results in qualitatively and quantitatively different GC profiles, e.g., dimethyl pyrazine and beta-phenylethanol are emitted under both growth conditions, whereas trimethyl pyrazine, tetramethyl pyrazine, and beta-phenylethyl acetate appear when glucose is not present in the medium (Fig. 1). Another example is the addition of l-glucose to the media, which leads to significant less volatile emission compared with growth on d-glucose (Fiddaman and Rossall 1993, 1994). Addition of trehalose to the media resulted in the emission of volatiles from Pseudomonas monteilii which stimulate mycelium growth of Pisollithus albus (Duponnois and Kisa 2006). Furthermore, bacterial metabolism varies in the lag, log, and stationary phases of S. rhizophila P69 batch cultures resulting in different emission patterns of, e.g., 2-piperidone, beta-phenylethanol, dimethyl pyrazine, trimethyl pyrazine, compound retention index 818 during the course of 144 h (Fig. 2). To assess temporal variations of volatile emission, variations that can contain important information for the detection and differentiation of microorganism. Online monitoring of characteristic emission patterns were recently performed with distinct volatiles released from Salmonella enterica and E. coli (Bunge et al. 2008).

Fig. 1
figure 1

VOC emission profiles of S. rhizophila P69. Headspace volatiles of bacteria grown on nutrient broth with 2% glucose (a) and without (b) were collected on super Q for 14 h beginning 58 h after inoculation. Volatiles were eluted from the absorbance material and analyzed by GC-MS. Individual compounds were identified: A2 = B3 (compound 2 of a = compound 3 of b): dimethyl pyrazine (93%), A5: trimethyl pyrazine (94%), A7: tetramethyl pyrazine (92%), A8 = B7: beta-phenylethanol (98%), A12: beta-phenylethyl acetate (90%). (%): identity to compound of the reference data library (NIST140)

Full size image
Fig. 2
figure 2

Emission of volatiles from S. rhizophila P69 in various growth intervals of batch culture. S. rhizophila was grown on nutrient broth with 2% glucose (a) and without (b), and volatiles were trapped on super Q in 5 or 14-h intervals. The time course of the unknown compounds with the retention indices (RI) 818 (black rhombus), 1124 (black square), 922 (black dot), 1264 (black triangle), 1008 (gray square), 1090 (gray dot), and 748 (a) (only 34 h, cross), 1189 (b) (only 144 h, black bar), 1136 (b) (only at 144 h, gray rhombus) are depicted. Insert: growth curves of batch cultures: optical density (OD, triangles) and living cell number (squares)

Full size image

Although temporal patterns and retention index assignments are helpful, a future task will be to elucidate the chemical structures of unidentified volatile compounds, since structural information will allow insights into the underlying metabolic pathway(s) that are active during bacterial growth. A goal for future research is to unravel the identity of the respective synthesizing enzymes and their specific regulation(s) in metabolically active bacteria. To date, only the biosynthesis of geosmin (trans-1,10-dimethyl-trans-9-decalol) responsible for the “earthy” and “musty” odor of many Streptomyces species is known (Bentley and Meganathan 1981; Cane and Watt 2003; Gust et al. 2003). Geosmin has been widely investigated because of its major contribution to the “off flavor” of contaminated drinking water, wines, and other foodstuffs, but it also contributes to the palatability of red beet and whisky aroma. Streptomyces species utilize both the mevalonate and non-mevalonate pathway, of which most of the enzymes had been discovered, to synthesize the terpenoid intermediate isopentenyl pyrophosphate (Takagi et al. 2000; Kuzuyama et al. 2000). Furthermore, sesquiterpene synthase genes have been isolated, which catalyze the cyclization reaction of farnesyl pyrophosphate (Gust et al. 2003). The increasing numbers of sequenced bacterial genomes, including bacteria that produce large amounts or a large diversity of volatiles, will support the elucidation of the biosynthetic pathways. Interest is presently focused on bacteria that emit volatiles with biological relevance, since the isolation of new bioactive natural compounds remains interesting for man because such compounds can be used, for example, to improve human health or to increase the productivity of agricultural products.

Functions of bacterial volatiles

The diversity of bacterial volatiles has a comparable complexity to that known for plants and fungi, and bacterial volatiles have turned out to be a rich source for new natural compounds. Presently, the biological functions of many bacterial volatiles are not understood in detail. Bacterial volatiles can be assumed to be similar to other volatiles and probably serve as (1) infochemicals for inter- and intraorganismic communication, (2) cell-to-cell communication signals, (3) a possible carbon release valve, or (4) growth-promoting or inhibiting agents. They are important for the sustainment of bacterial populations in ecological niches and for the cooperative development of a community of different organisms, and they can support the selective advantage of some community member(s) and survival during evolution.

Volatiles are important chemicals because they can act over a wide range of scales. Their action profile ranges from the ability to diffuse through aqueous solutions to being able to permeate through the atmosphere. Therefore, volatiles not only play a role above ground but also function below ground.

Bacterial volatiles effecting fungi

An early report on bacterial volatiles affecting fungal growth and development was published by McCain (1966) showing that Streptomyces griseus volatiles reduced the sporulation of Gleosporium aridum and induced the formation of sclerotia in Sclerotium cepivorum and Rhizoctonia solani. The effects of bacterial volatiles on fungi known to date range from the stimulation of fruit body formation, to spore germination, to mycelium growth inhibition and promotion, and to the stimulations or reduction of sporulation (Table 1). A large survey by Wheatley (2002) has revealed that 250 bacterial soil isolates cause fungal micelle growth inhibition or promotion, with the two most active bacteria being Citrobacter freundii and Pseudomonas fluorescens. In a recent investigation of 1,018 randomly selected bacteria, 32% have turned out to produce fungistatic volatiles (Zou et al. 2007). The producers of bioactive volatiles are species of Alcaligenes, Bacillus, Ensifer, Lysobacter, Planomicrobium, Sporosarcina, and Stenotrophomonas. Other investigations have named Burkholderia, Pseudomonas, Serratia, Xanthomonas, Pectobacterium, and Agrobacterium species as potent antifungal bacteria (Table 1). The volatiles from any one bacterial strain do not cause the same effects or the same degree of inhibition to all the fungi; rather, the responses depend on the specific fungi–bacteria combination (Tables 1 and 2). The following reasons might be responsible for these differences: (1) different fungi may respond to different component(s) of the volatile mixture, or (2) the sites of action may be different, or (3) the fungi might possess different abilities to detoxify the volatile metabolite(s).

Table 1 Effects of bacterial volatiles on fungi
Full size table
Table 2 Mycelium growth inhibition of phytopathogenic fungi by volatiles of phytopathogenic bacteria
Full size table

Volatiles of rhizobacteria and of phytopathogenic bacteria exert fungistatic effects, for example on the growth of phytopathogenic fungi such as Verticillium dahliae, Sclerotinia sclerotiorum, and R. solani. This adds an additional facet to the antagonistic potential of plant-growth-promoting bacteria and is of agricultural interest (Table 2). Indeed, some bacteria prevent symptom development in field rape plants (Alstrom 2001). Other studies have indicated that the promotion of mycorrhizal formation, e.g., by Glomus mosseae, is due to the presence of rhizobacteria or Streptomyces species (Azcon-Aguiler and Barea 1985; Azcon-Aguiler et al. 1986; Fitter and Garbaye 1994). Although Streptomyces spp. are important positive players in arbuscular mycorrhiza formation, it has to be proven that volatiles are key components (Schrey et al. 2005).

Microscopic observations of the fungal mycelium during Bacillus subtilis volatile fumigation have revealed effects on hyphae and conidia (Chaurasia et al. 2005). The longitudinal and traverse septae completely disappeared in Alternaria alternata, and conidia became thick-walled and spherical or irregular in shape. Upon volatile exposition, Cladosporium oxysporum conidiophores became vegetative and stunted. Lysis of fungal hyphae, vacuolization, and granulation in mycelium structures have been observed in Fusarium oxysporum and Phytium afertile. Further studies need to be performed in order to understand the underlying mechanisms that lead to these morphological aberrations after exposure to bacterial volatiles.

In addition to the observed morphologic changes, alterations at the molecular level are expected in the fungi. In Phanaerochaete magnoliae, enzyme activities change on bacterial volatile exposure, e.g., laccase activity ceases completely, whereas tyrosinase activity increases, both probably as a result of the up- and down-regulation of gene expression rather than of direct enzyme activity changes (Mackie and Wheatley 1998). In several cases, we have observed the dark coloration of the growth media attributable to melanin production of the fumigated fungi (not shown, Kai /Piechulla, unpublished results).

The fungistatic effect attributable to bacterial volatiles has been substantiated during the past 10 years. Volatile mixtures emitted from bacteria have often been applied, but the bioactive compound(s) remain to be determined. For example, the mycelium growth of Fusarium culmorum shows a dose-dependent inhibition by dimethyl disulfide, and small effects have been obtained with 1-undecene (Fig. 3). Additional bacterial compounds that act as bioactive agents on fungi include various amines, benzaldehyde, cyclohexanol, decanal, 2-ethyl-1-hexanol, nonanal, benzothiazole, and dimethyl trisulfide; compounds such as methyl pyrazine, 2,5-dimethyl-pyrazine, phenylenediamine, 4-octylbenzoic acid, several middle and long chain alkans, alkenes, aldehydes, and alcohols have surprisingly low inhibitory effects on various fungi (Table 1). Dimethyl disulfide and benzaldehyde possess inhibitory but sometimes non-inhibitory effects; this might be due to different experimental setups, e.g., the way that the compounds are applied and their concentrations. Since the GC profiles of bacterial volatiles allude to many more compounds, future work has to concentrate on the structural elucidation and on the determination of biological activity.

Fig. 3
figure 3

Mycelium growth of F. culmorum exposed to dimethyl disulfide (DMDS) or 1-undecene. Bipartite Petri dishes were inoculated with F. culmorum on one side. Filter paper with pentane (a), 5.2 μg DMDS in pentane (b), 0.52 μg DMDS in pentane (c), and 3.74 μg 1-undecene in pentane (d) was applied on the other side (incubation time, 4 days)

Full size image

Bacterial volatiles effecting plants

The first (and so far only) report on the growth promotion of Arabidopsis thaliana attributable to two typical VOCs emitted by bacilli, 2,3-butandiol and acetoin, was published by Ryu et al. (2003), whereas the biological functions of the other 38 compounds, comprising short chain alcohols, aldehydes, acids, esters, ketones, hydrocarbons, S-containing compounds, and CO2 that are emitted from B. subtilis and B. amyloliquefaciens remain to be elucidated (Farag et al. 2006). In contrast, strong inhibitory effects on A. thaliana and the moss Physcomitrella patens have been demonstrated by other bacteria, such as Pseudomonas spp., Serratia spp., and Stenotrophomonas spp., although the bioactive compounds are presently unknown (Table 3). To our knowledge, bacterial volatiles have so far not been tested on agriculturally relevant plant species. The scientific advantage of studying A. thaliana, however, is that whole genome microarray analysis can be performed. Out of 26,000 protein-encoding transcripts, 600 differentially expressed genes related to cell wall modifications, primary and secondary metabolism, stress responses, and hormone regulation have been identified in A. thaliana exposed to B. subtilis volatiles (Zhang et al. 2007). These data implicate the regulation of auxin homeostasis and cell expansion and provide a new paradigm as to the way that bacilli promote plant growth.

Table 3 Effects of bacterial volatiles on plants
Full size table

Bacterial volatiles effecting animals

Since rhizobacterial volatiles influence the growth of fungi and plants, the question has arisen as to whether soil-living animals are also affected (Table 4). The protozoan Acanthamoeba castellanii und Paramecium caudatum are dominant soil organisms (1.6 × 105 individuals per gram soil) primarily feeding on bacteria (Bonkowski 2004). Co-cultivations in which organisms have direct contact have shown that E. coli and Stenotrophomonas maltophilia serve as a good food source for A. castellanii, whereas Staphylococcus epidermidis, S. marcescens, and Pseudomonas aeruginosa are not feeder bacteria (Wang and Ahearn 1997). Although not investigated by the authors, it cannot be excluded that the bacterial volatiles have a direct impact on the feeding behavior of A. castellanii. We have shown that in bipartite Petri dishes in which the different organisms never come into direct contact, the volatiles of B. subtilis, P. fluorescens, S. odorifera, and Xanthomonas campestris pv vesicatoria negatively influence the growth of A. castellanii and P. caudatum. Although the inhibition of A. castellani after 4 days of co-cultivation ranges between 60% and 95%, depending on the density of the protozoa, the bacterial volatiles are lethal for P. caudatum at all population densities (Fig. 4).

Fig. 4
figure 4

Co-cultivation of bacteria with A. castellanii. S. odorifera 4R × 13 (light gray bar; CFU/ml, 1 × 109), B. subtilis B2g (dark grey bar; CFU/ml, 8.6 × 107), P. fluorescens L13-6-12 (white bar; CFU/ml, 6.1 × 108), and Xanthomonas vesicatoria pv. vesicatoria 85-10 (middle grey bar; CFU/ml, 1.9 × 108) were co-cultivated in bipartite Petri dishes with A. castellanii at three population densities (2.6 × 103, 2.6 × 104, 2.6 × 105 cells/ml). Black bar: control with water

Full size image
Table 4 Effects of bacterial volatiles on animals
Full size table

Other prominent soil creatures are nematodes such as the non-parasitic bacterial feeding Caenorhabditis elegans. C. elegans lives in the soil at the air–water interface; it therefore can encounter both water-soluble and volatile chemicals that might influence its behavior. Pseudomonas and other microbes produce small chain alcohols, ketones, diacetyl, and esters as metabolic byproducts (Zechman and Labows 1985; Dainty et al. 1985) that can act as natural chemoattractants for C. elegans (Bargmann et al. 1993). Indeed, C. elegans provides two types of chemosensory neurons that can detect volatile attractants (Sengupta et al. 1996). Each pair of neurons detects several different odorants: One set of olfactory neurons senses diacetyl, pyrazine, and thiazols, whereas the other set is activated by benzaldehyde, butanone, isoamyl alcohol, and thiazoles. Our co-cultivation experiments of C. elegans with B. subtilis, P. fluorescens, S. odorifera, and X. campestris pv vesicatoria in bipartite Petri dishes have substantiated previous observations. The volatiles emitted by these bacteria are so attractive for this worm that they manage to crawl over a 3-cm-high barrier, most likely to establish a new feeding source (Molina/Piechulla, unpublished results). In contrast, the free-living nematode Panagrellus redivivus and the pinewood nematode Bursaphelenchus xylophilus gradually reduce their motility (or die) during exposure to bacterial volatiles such as phenol, octanol, benzaldehyde, benzene acetaldehyde, decanal, 2-nonanone, 2-decanone, cyclohexene, and dimethyl disulfide (Gu et al. 2007). In summary, nematode susceptibility to one or another compound varies, indicating a species-specific reaction to individual compounds; therefore, certain bacterial isolates have the potential to be effective biocontrol agents against nematodes.

Recently, gravid mosquito (Aedes aegypti) females have been shown to use bacterial (alpha and gamma proteobacteria) volatiles, e.g., carboxylic acids and methyl esters, as potent oviposition stimulants to direct egg laying in favorable habitats (Ponnusamy et al. 2008). Another example of bacterial odor-dependent oviposition has been reported for gravid Anopheles gambiae (Huang et al. 2006). Such knowledge of bioactive bacterial volatiles/volatiles mixtures and their respective stimulative concentrations can be used to study insect epidemiology and might have an impact on mosquito-abatement programs.

The aforementioned examples indicate that bacterial volatiles affect invertebrate animal growth and behavior. The effects of bacterial volatiles on vertebrate animals or respective cell cultures has been reported by Kurita-Ochiai et al. (1995) who have shown that fatty acids (e.g., butyric acid, propionic acid, valeric acid, and isovaleric acid) produced by periodontophatic bacteria, such as Porphyromonas gingivalis, Prevotella loescheii, and Fusobacterium nucleatum, impair lymphocyte proliferation and cytokine production.

Bacterial volatiles effecting bacteria

Bacteria use a cell-to-cell communication system to monitor their population density (quorum sensing; reviewed by Ryan and Dow 2008). This allows a colony or a group of organisms to behave in a coordinated fashion in order to regulate processes contributing to virulence, antibiotic production, biofilm formation, and other developmental programs (Table 5). Additionally, bacteria can sense signal molecules that they do not synthesize, thereby eavesdropping on signaling by other organisms in their immediate environment. The autoinducer signal molecules produced by bacteria are structurally diverse; however, many are compounds with small molecular masses that might have the potential to act as a volatile. Gram-negative bacteria use N-acylhomoserine lactones (N-AHLs), fatty acid derivatives (3-hydroxypalmitic acid methyl ester, cis-unsaturated fatty acids), and cyclic dipeptides, whereas Gram-positive bacteria use amino acids and modified peptides or gamma-butyrolactones (Streptomyces spp.) for communication. Other known signals are methyl-2,3,3,4-tetrahydroxyhydrofuran (S-THMF), indole, quinolones, and S-3-hydroxytridecan-4-one. The production of indole is widespread among soil bacteria and is generated through the degradation of tryptophan by tryptophanase. Indole has been shown to regulate the expression of several multi-drug reporter genes in E. coli via two two-component signal transduction pathways; it also inhibits biofilm formation in E. coli, P. fluorescens, and P. aeruginosa. A butyrolactone (diffusible factor) has been shown to regulate the production of the yellow pigments (xanthomonadins) in X. campestris. During our investigations, we have observed that the typical red colony coloration of B. cepacia diminishes on exposure to volatiles from S. odorifera and Serratia plymuthica (Fig. 5), showing that volatiles of bacteria can influence the metabolism of other bacteria. Another example is the release of volatile short chain fatty acids from Veilonella species and Bacteroides fragilis, which thus control the growth of the enteropathogens Salmonella typhimurium, Salmonella enteritidis, E. coli, and Pseudomonas aeroginosa (Hinton and Hume 1995; Wrigley 2004). The acids reduce the pH milieu of the intestine, resulting in a higher sporulation rate of Clostridium difficile or Clostridium perfringenes. These are conditions under which such anaerobic bacteria produce their toxins that finally cause diarrhea in the patient. This example shows that communication via bacterial volatiles is also of significant clinical relevance.

Fig. 5
figure 5

Co-cultivation of B. cepacia and S. odorifera. B. cepacia 1S18 (left sector) cultivated with (a) non-inoculated nutrient broth II (right sector) or (b) with S. plymuthica HRO-C48 (right sector). Discoloration of B. cepacia colonies is visible after 72 h of co-cultivation

Full size image
Table 5 Effects of bacterial volatiles on bacteria
Full size table

Conclusions and outlook

This review was intended to summarize the present knowledge of bacterial volatile emission as well as the effects of the volatiles exert on different organisms. The inquest clearly demonstrates that the diversity and complexity of bacterial volatile emanation is similar to other organisms and that it was underestimated in the past. Two major tasks should be addressed in the future. Firstly, structural elucidation of new natural products with specific action potentials would increase the impact of biologically active compounds, and secondly, functional analysis would clarify the role of volatiles in the interactions of organisms, populations, and communities and reveal their importance for the ecological balance.