Introduction

Imidazolium compounds widely exist in nature. Imidazole is not only part of the amino acid histidine, but is also used in the synthesis of a wide range of antifungal agents, e.g., clotrimazole, econazole, imazalil. 2-Methylimidazole is used for the production of metronidazole, an agent used for the treatment of trichomoniasis and amoebiasis. Lepidilines B (1,3-dibenzyl-2,4,5-trimethylimidazolium chloride), isolated from the root extract of Lepidium meyenii, exhibits micromolar cytotoxicity against several human cancer cell lines (Cui et al. 2003). This alkaloid is synthesized (Wolkenberg et al. 2004) by preparation of the intermediate 2,4,5-trimethylimidazole (TMI) (Fig. 1, left).

Fig. 1
figure 1

Chemical structures of 2,4,5-trimethylimidazole (TMI) and 2,3,5,6-tetramethylpyrazine (TTMP)

Full size image

2,3,5,6-Tetramethylpyrazine (TTMP) (Fig. 1), a biologically active ingredient originally isolated from the rhizome of Ligusticum wallichii, is one of the main pyrazines detected in many cocoa bean- or soybean-based fermented foods (Hashim et al. 1997; Kosuge et al. 1971; Zak et al. 1972). Besides TTMP, 2,5-dimethylpyrazine, 2,3,5-trimethylpyrazine, and 2-ethyl-3,5,6-trimethylpyrazine are other fermentation products produced by Bacillus strains (Besson et al. 1997; Larroche et al. 1999; Xiao et al. 2006). Although structurally similar to pyrazines, alkyl imidazolium compounds have been rarely reported during bacterial fermentation processes.

In this study, the formation of TMI is demonstrated for the first time during TTMP production from glucose by Bacillus strains. The mechanism and key factors affecting the production of this alkaloid are discussed.

Materials and methods

Organisms and growth

Bacillus sp. RX3-17, a mutant for TTMP production (Xiao et al. 2006), and Bacillus subtilis 10025, a strain for acetoin production (Xiao et al. 2007), have been used in batch cultures carried out in a 5 l fermenter (Biostat B, B. Braun Biotech International GmbH).

Analytical methods

Cell growth was measured at 620 nm and converted to the cell dry weight (CDW); OD620nm of 1 = 0.31 g CDW l−1. Glucose was enzymatically determined using YSI 2700 Select biochemical analyzer (Yellow Springs Instrument Co.). Products in the culture broth were extracted with dichloromethane and analyzed using a GC system equipped with a flame ionization detector and a 30 m SPB-5 capillary column (Supelco). The operational conditions were similar to those used for TTMP determination (Xiao et al. 2006). Analysis of unknown products occurred in GC determination was performed by a Saturn 2000 GC-MS system (Varian) equipped with a 30 m DB-5MS column (J & W Scientific). The compounds were tentatively identified by comparing their mass spectra with those contained in the MS data system (Saturn library). GC—high resolution MS (GC-HRMS) was performed on an Agilent 6890 GC system equipped with a DB-5MS column and a TOF-MS (GCT, Waters). Mass spectra in the electron impact mode were both generated at 70 eV. For amino acid analysis, samples were centrifuged at 10,000×g for 5 min at 4°C. The supernatants were treated with trichloroacetic acid (100%, w/v) at 4°C overnight before centrifugation again at 10,000×g for 15 min. The final supernatants were appropriately diluted and then analyzed with an amino acid analyzer (Hitachi 835-0200).

Results and discussion

Production and identification of TMI

During TTMP production from glucose by Bacillus sp. RX3-17, an unknown compound was detected by GC after glucose was depleted in the culture medium (Fig. 2a). GC-MS determination of this compound only met an moderate match with 1,3,5-trimethylpyrazole in the MS data library. Its mass spectrum had a characteristic feature of alkyl imidazolium compounds, e.g., 2,5-dimethylimidazole and 2-ethyl-5-methylimidazole. Considering the chemical structure of TTMP, the unknown compound was speculated to be TMI. For further identification, TMI was chemically synthesized (Wolkenberg et al. 2004; Harper et al. 1997) and there was a perfect GC-HRMS match, including both GC retention times and HRMS profiles (Fig. 3), to conclude that the unknown compound was TMI.

Fig. 2
figure 2

TMI production during TTMP batch fermentation in a 5-l bioreactor with a Bacillus sp. RX3-17 and b B. subtilis 10025, using the medium containing (g l−1): glucose, 80; yeast extract, 50; peptone, 10; and (NH4)2SO4, 30. Inverted triangle, glucose; square, CDW; asterisk, acetoin; triangle, TTMP; circle, TMI. Batch culturing was conducted with stirring at 700 rpm, air flow at 1.0 vvm, and temperature at 37°C. The pH was controlled at 7.0 before transition into stationary phase, and then 7.7

Full size image
Fig. 3
figure 3

TMI identification by GC-HRMS comparative analysis. Bio-TMI and Chem-TMI samples were extracted from culture media and chemical reaction mixtures, respectively. Asterisks indicate the GC peaks of TMI. HRMS profiles refer to the corresponding peaks with asterisks

Full size image

Since Bacillus sp. RX3-17 was a mutant strain, was the occurrence of TMI in the fermentation broth just a coincidence? B. subtilis 10025 was examined. A similar time course of the formation of TMI was recorded (Fig. 2b), indicating that the occurrence of this alkaloid was not produced accidentally in the culture medium. To our best knowledge, this is the first discovery of TMI in microbes. As indicated previously, imidazolium compounds are often biologically active and used as antifungal agents. Thus, the occurrence of TMI during Bacillus cultivation may be related to antibiotic activities.

Formation mechanism

What are the precursors of TMI during bacterial fermentation? l-Threonine and acetoin occur as precursors of 2,5-dimethylpyrazine and TTMP, respectively (Besson et al. 1997). Larroche et al. (1999) proposed the biosynthetic pathways for 2,5-dimethylpyrazine and TTMP following the previous studies by others (Miller et al. 1973; Kim et al. 1994). Therefore, amino acids and acetoin are possible precursors of TMI.

Sixteen amino acids occurred during Bacillus sp. RX3-17 batch fermentation (Fig. 2a) and their concentrations were evaluated, as shown in Supplementary Table 1. These gave no obvious clue about the kinetics of TMI formation. Nevertheless, Bacillus strains are good acetoin producers. The metabolism of acetoin in bacteria is strictly regulated by glucose (Xiao and Xu 2007). The results shown in Fig. 2 reveal that Bacillus sp. RX3-17 and B. subtilis 10025 share similar common processes of TMI production: at the time point of glucose depletion in the culture medium, acetoin began to decrease whilst TMI began to appear and then grew in concentration. But acetoin is not likely be the precursor of TMI because, as time went on, the level of acetoin in the culture medium became constant whilst the concentration of TMI steadily increased. Further studies are under way to elucidate the mechanism of TMI formation in bacteria.

Influence of ammonium salt on TMI production

(NH4)2HPO4 was better than (NH4)2SO4 in TTMP production by Bacillus sp. RX3-17 (Xiao et al. 2006). Only about half amount of TTMP was produced when (NH4)2HPO4 was replaced by (NH4)2SO4 as ammonium source. However, for TMI production in this study, only 0.02 to 0.04 g TMI l−1 was detected when (NH4)2HPO4 was used to replace (NH4)2SO4. The difference clearly indicated that the choice of ammonium salts could alter the relative amount of TTMP and TMI, but the question why (NH4)2SO4 was much more favorable than (NH4)2HPO4 remains unresolved.