The isolation, purification and biological activity of a novel antibacterial compound produced by Pseudomonas stutzeri


  • Editor: Bruce Ward

Correspondence: David H. Edwards, Department of Molecular and Cellular pathology, University of Dundee, Ninewells Hospital and Medical School, Dundee, UK. Tel.:+44 (0) 1382 496388; fax:+44 (0) 1382 225163; e-mail:


A novel compound designated zafrin [4β-methyl-5, 6, 7, 8 tetrahydro-1 (4β-H)-phenanthrenone] was isolated from a crude extract of a marine bacterium identified as Pseudomonas stutzeri. Zafrin showed strong antibacterial activity against both Gram-positive and Gram-negative bacteria. The compound was purified and its structure was elucidated by spectroscopic methods including 1H-nuclear magnetic resonance (NMR), 13C-NMR, 1D-NMR and 2D-NMR spectroscopy. It could be demonstrated that a purified solution of zafrin was active against several human pathogens, including Staphylococcus aureus, and Salmonella typhi. By contrast, zafrin did not inhibit the growth of eukaryotic organisms Candida albicans and Schizosaccharomyces pombe. The minimal inhibitory concentration for Gram-positive bacteria ranged from 50 to 75 μg mL−1 and varied between 75 and 125 μg mL−1 for Gram-negative bacteria. Zafrin lysed Bacillus subtilis cells grown in an osmotically protected medium, suggesting that it does not act upon the cell wall. Further investigation using B. subtilis indicated that the compound is bactericidal and is likely to target the cell membrane.


Newly emerging infectious diseases, re-emergent diseases and multidrug-resistant bacteria mean that there is a continued need to develop new antibiotics. The sea is an immense and practically unexploited source of potentially useful biologically active substances. Chemical compounds of marine microorganisms are not as well characterized as those of their terrestrial counterparts and marine bacteria have recently been identified as a source of several new bioactive metabolites (Bernan et al., 1997; Jaruchoktaweechai et al., 2000). Interactions between microorganisms are a well-known phenomena, and substrate competition and antagonism are important factors in the selection of microbial communities in any given ecological niche (Fredrickson & Stephanopoulos, 1981). The competitive or antagonistic activity of Pseudomonas sp. in the rhizosphere has been very well described (Casida, 1992) and the antimicrobial activity of Pseudomonas sp. has been recognized as a major factor in the suppression of many root pathogens. A few studies have also described the antibacterial effect of aquatic Pseudomonas sp. (Gram, 1993) but rarely the has inhibitory mechanism been elucidated (Wratten et al., 1977).

One of the major problems that health care is currently facing is the emergence of antibiotic-resistant bacteria and in particular the spread of methicillin-resistant strains of the pathogenic bacteria Staphylococcus aureus (MRSA). MRSA is now resistant to almost all commonly used antibiotics, and even resistance to one of the ‘last-choice’ antibiotics, vancomycin, has begun to appear (Courvalin, 2006; Deurenberg et al., 2007). In this context, a screening program was undertaken for antibacterial substance production by marine bacteria. One of the isolated strains, CMG1030, showed high levels of antibacterial activity, and was identified as belonging to the species Pseudomonas stutzeri. In this study, the antibacterial substance produced by CMG1030 was purified, its chemical structure was determined and its antibacterial and bactericidal activities against Gram-negative and Gram-positive bacteria were investigated.

Materials and methods

Bacterial strains and media used

Clinical isolates belonging to the species Staphylococcus aureus (MRSA), Staphylococcus epidermidis, Corynebacterium xerosis, Pseudomonas aeruginosa, Escherichia coli and Proteus mirabilis were used to evaluate the antibacterial activity of the isolated compound (Karachi Hospital). In addition, the reference strains Bacillus subtilis 168 and Salmonella typhi ATCC13311 were also used. All of the bacterial strains were kept as stock cultures in trypticase soy broth (TSB medium) (Oxoid) containing 20% (v/v) glycerol at −70 °C. The standard strains of Candida albicans and Schizosaccharomyces pombe were also used in this work.

Screening for antibacterial activity

Screening of bacterial isolates from fish gut samples was performed essentially as described previously (Uzair et al., 2006). Importantly, the authors concentrated on a selection of over 100 morphologically distinct colonies, identified during enrichment, that had the ability to inhibit the growth of the clinical strain of MRSA. The isolate responsible for the greatest zone of inhibition was chosen for further study.

Isolation of antibacterial compound

Zafrin was isolated from strain CMG1030 grown to confluent growth on 100 King B (King et al., 1954) agar plates (90 mm) incubated at 30 °C for 5 days. The 250 mL of agar was cut into 1-cm squares and extracted three times with 80% (v/v) acetone in water. The combined extracts were filtered through cheesecloth to remove pieces of agar from 100 plates; other particulate matter was removed by centrifugation at 9000 g for 15 min at 4 °C. The supernatant was collected and evaporated under a vacuum to remove the acetone. NaCl (5 g) was dissolved in each 100 mL fraction of the aqueous concentrate and then extracted with ethyl acetate. The final ethyl acetate extracts were combined and evaporated under vacuum to yield 5 g of material. The crude extract was subjected to column chromatography (CC) using a 6 cm by 70 cm silica gel column (70–230 mesh, E. Merck) and a linear solvent gradient of 5–100%n-hexane and ethyl acetate. A total solvent volume of c. 2.5 L was applied, and 95 25 mL fractions were collected and analyzed by High-performance thin layer chromatography (HPTLC). HPTLC was performed with precoated silica gel G25-UV254 plates and detection was performed at 254 nm using ceric sulfate spray reagents. In total, 30 fractions (32–62) contained one of three UV active spots and these were pooled for a second round of purification. For the second round of CC, a 2 cm by 8 cm column, 230–400 mesh flash silica gel and a linear solvent gradient of 5–70%n-hexane and ethyl acetate (2.3 L) were used. A single UV active compound was detected by HPTLC in 29 (6 mL) fractions that were eluted using n-hexane and ethyl acetate (7 : 3). The fractions were pooled and the solvent was evaporated under vacuum to produce 6 mg of a yellow powder with antimicrobial activity.

Nuclear magnetic resonance (NMR)

Optical rotation was measured on a DIP-360, Digital polarimeter. The infrared spectra were recorded in KBr on a Jasco-320 A spectrometer. The UV spectrum was scanned on Hitachi-UV-3200. Mass spectra (EI and HR-EI-MS) were measured in an electron impact mode of a Finnigan MAT312 spectrometer with ions given in m/z (%). 1H-NMR spectra were obtained on a Bruker AM-400 spectrometer with tetramethysilane (TMS) as an external standard in CDCl3. 13C-NMR, COSY, HMBC and HMQC spectra were measured in CDCl3 on a 400 MHz Bruker AM-400 spectrometer. 2D NMR spectra were recorded on a Bruker AMX 500 NMR spectrometer. The DEPT experiments were carried out at 45, 90 and 135 °C.

Determination of minimal inhibitory concentration (MIC)

The minimal inhibitory concentration of the antibacterial compound was initially determined by a disc diffusion method (Uzair et al., 2006), the MIC representing the minimum concentration that produced a 10 cm, or greater, zone of inhibition. These values were then checked in liquid media using the serial dilution method and the lowest value recorded (Table 1).

Table 1. In vitro activity of Zafrin against reference microorganisms
OrganismZone of
inhibition (cm)*
(μg mL−1)
  • *

    Zone of inhibition produced by Zafrin at the appropriate MIC reported in column 3.

  • R, resistance.

Staphylococcus aureus (MRSA)3160
Staphylococcus epidermidis1550
Bacillus subtilis2850
Corynebacterium xerosis2875
Salmonella typhi21100
Pseudomonas aeruginosa*R*R
Escherichia coli20100
Proteus mirabilis11125
Candida albicans*R*R
Schizosaccharomyces pombe*R*R

Time kill experiment

The MIC for each antibiotic was determined in preliminary experiments. Briefly, exponentially growing B. subtilis cells were used to inoculate a series of tubes containing 2 mL of nutrient broth (Oxoid) (A600 nm 0.05) and a set concentrations of antibiotic. The cultures were grown overnight at 37 °C without shaking, and the presence of viable cells was determined by assaying 200 μL of each culture for the ability to generate colonies on nutrient agar plates. The time-kill experiments were performed in triplicate, and essentially as described by Aeschlimann & Rybak (1998). Each experiment was conducted in 500-mL Erlenmeyer flasks containing 100 mL of fresh nutrient broth inoculated with exponentially growing cells of B. subtilis (A600 nm 0.4). Inoculation was carried out immediately after the addition of the antibiotic at a final concentration of 1/4, 1/2, 1, 2 or 4 times the MIC. The flasks were incubated for 24 h at 30 °C and the viable cell count was determined at various time points by the plating of serial dilutions on nutrient agar (37 °C for 24 h). To minimize the effect of antibiotic carryover, the samples were centrifuged at 5000 g for 2 min before plating out and the antibiotic medium was replaced with fresh nutrient broth.

Investigation of zafrin's mode of action

The ability of zafrin and known cell wall-active antibiotics to induce the formation of osmotically fragile spheroplasts was tested on an exponentially growing culture of B. subtilis essentially as described by Singh et al. (2003). To determine whether the cell wall was the site of zafrin action, osmotically protected B. subtilis protoplasts were generated by lysozyme treatment in the presence of 0.5 M sucrose (Perry & Edwards, 2006). The protoplasts were then incubated for 10 min in the presence of zafrin (50 μg mL−1), and a sample was observed by phase contrast microscopy.

Phase contrast microscopy

All microscopic images were obtained by imaging exponentially growing cultures of either Staphylococcus aureus or B. subtilis treated with the appropriate antibiotic. Time lapse images of exponentially growing cells treated with an antibiotic were obtained using agarose-covered slides (Edwards & Errington, 1997). Bacterial growth was followed for up to 60 min, although typically images of lysing bacteria were obtained during the first 10–30 min of exposure. The microscope image data were handled as described previously (Perry & Edwards, 2004).


Identification of an antibiotic-producing organism

A collection of more than 100 bacteria isolated from the intestinal tract of fish that landed on the Baluchistan coast that borders the Gulf of Karachi (Pakistan) was screened. One isolate (CMG1030) obtained from the intestinal tract of a Ribbon fish (Desmodema spp.) showed good inhibitory activity against several species of bacteria, including the MRSA strain (Fig. 1). Biochemical analysis and subsequent 16S rRNA gene typing identified this isolate to be Pseudomonas stutzeri, an organism that is widely distributed in a range of natural environments and noted for its metabolic diversity (Lalucat et al., 2006). A confluent, 5-day-old, culture of CMG1030 grown on King B medium appeared to produce significant amounts of the antibacterial compound. Analysis of CMG1030 established that antibacterial activity could be detected in an ethyl acetate fraction of crude cell extract, and this suggested that the substance was not bound to the cell surface. The ethyl acetate extract was then purified using CC and a final set of fractions was assayed for antimicrobial activity. In total, 29 fractions were found to have an inhibitory effect on MRSA and to contain a single compound identified by TLC. This single compound was named zafrin and was the subject of further analysis.

Figure 1.

 Cross-streak analysis of organisms sensitive to the antibiotic-producing strain CMG1030. A sample taken from a 48-hour-old stationary phase culture of CMG1030 (A600 2.6) was streaked vertically on a Muller Hinton agar plate and then incubated at 30°C for 24 h. To kill the antibiotic-producing strain, the plate was inverted and a 9 cm2 Whatman filter paper (3 M) was impregnated with 0.3 mL of chloroform placed inside the lid. The lid was then replaced and the chloroform filter paper was left in place for 30 min, before being discarded and a new lid was applied to the Petri dish. The five test organisms, (A) Staphylococcus aureus MRSA, (B) Staphylococcus epidermidis, (C) Enterococcus faecalis, (D) Bacillus subtilis, (E) Salmonella typhi, were then horizontally cross-streaked and the plate was incubated at 37°C for 24 h. The absence of growth across the center of the plate illustrates the sensitivity of the reference organisms.

Chemical structure of zafrin

Following isolation of zafrin, the compound was analyzed using 1H-NMR (CDCl3, 500.23 MHz), 13C-NMR (CDCl3, 100.61 MHz), high-resolution electron ionization mass spectral (HREIMS), DEPT and 1H-1H COSY. This identified zafrin to be 4β-methyl-5, 6, 7, 8 tetrahydro-1(4β-H)-phenanthrenone (supplementary Appendix S1 and Table S1). The molecular formula, C15H16O, of zafrin was deduced and the resulting structure (Fig. 2) suggested that zafrin would be a stable uncharged compound that is likely to be hydrophobic and lipophilic.

Figure 2.

 Chemical structure of zafrin.

Antibacterial activity

Zafrin inhibited the growth of several major human pathogens, including Staphylococcus aureus, Corynebacterium xerosis, Salmonella typhi, Proteus mirabilis and E. coli (Table 1). However, it did not cause the death of Pseudomonas aeruginosa, suggesting that this closely related species has a natural resistance to the compound. In addition, zafrin did not kill Candida albicans or Schizosaccharomyces pombe, suggesting it is not toxic to single-celled eukaryotes and is likely to have a specific prokaryotic target. In general terms, the MIC of zafrin was lower for Gram-positive bacteria (50–75 μg mL−1) than for Gram-negative bacteria (75–125 μg mL−1). This remained true for eight more organisms that were tested (data not shown). To investigate the activity of zafrin further, a Gram-positive model microorganism (B. subtilis 168) was chosen. Bacillus subtilis is commonly used to screen and evaluate novel antimicrobial compounds (Hutter et al., 2004; Thomaides et al., 2007).

Zafrin activity on B. subtilis cells

To determine the efficiency of zafrin action, a series of experiments were conducted that compared the rate of zafrin killing with three other commonly used antibiotics. The time kill kinetics showed that zafrin was able to kill B. subtilis rapidly at its MIC of 50 μg mL1− (Fig. 3). The rate at which bacterial cell counts decreased was even more rapid when the culture was exposed to twice the MIC, and using four times the MIC it was observed that no viable cells could be recovered 6 h after exposure to the antibiotic. In comparison, the effect of vancomycin, ampicillin and tetracycline was more gradual, with a 100% reduction in viability requiring at least a 10-h exposure.

Figure 3.

 Killing kinetic curves for zafrin and reference antibiotics. Experiments performed in triplicate on exponentially growing cells of Bacillus subtilis 168 using MIC concentrations of zafrin (50 μg mL−1), ampicllin (10 μg mL−1), tetracycline (2 μg mL−1) and vancomycin (0.65 μg mL−1). ▪, control; ▴, 1/4MIC; •, 1/2MIC; *, MIC; -, 2MIC; _, 4MIC.

Microscopic analysis of either B. subtilis or Staphylococcus aureus cultures 30 min after the addition of zafrin revealed a large amount of cell debris and very few intact cells (Fig. 4a i and 4b). Unusually, the septa could be visualized in the majority of intact B. subtilis cells. Observing division sites in chains of B. subtilis cells is a difficult procedure due to the presence of a thick cell wall that is slow to split following cell division. To overcome this, and perform cell length analysis, cultures of B. subtilis were exposed to 70% ethanol, a membrane-disrupting agent (Ingram & Buttke, 1984), that causes the cytoplasm to contract away from the cell wall and reveal the location of the septum (Hauser & Errington, 1995). The similarity between the intact cells that were observed (Fig. 4a i) and ethanol-fixed cells (Fig. 4a ii) suggested that a similar process occurs to cells exposed to ‘sublytic’ levels of zafrin.

Figure 4.

 Phenotype of antibiotic-treated cells. (a) i, Phase contrast image of mid-exponential Bacillus subtilis cells 30 min after exposure to zafrin (50 μg mL−1). The arrows indicate where the cytoplasm of intact cells has shrunk to reveal the location of septa. (a) ii, Corresponding image of B. subtilis cells fixed with 70% ethanol. The arrow indicates a division site that would not normally be observed in cells taken directly from liquid culture. (b) Image of Staphylococcus aureus cells exposed to zafrin (60 μg mL−1). (c) and (d) Phase contrast images of B. subtilis cells 30 min after exposure to either 5 μg mL (c) or 50 μg mL−1 (d) nisin. Note that the ‘ghost’ cells produced by a low concentration of nisin (indicated by arrows) are not observed after the addition of a membrane-disrupting concentration of nisin (d). (e) Microscope images of B. subtilis exposed to zafrin (25 μg mL−1) and transferred to an agarose-covered slide. (e) i, cells photographed immediately after transfer and ii, a second image obtained 11 min later. Note the ‘phase-bright’ regions within the cell and the reduction in the cell's overall dimensions. (f) Corresponding experiment to (e) i and ii with Triton X-100. Bacillus subtilis cells exposed to 0.04% Triton X-100. (f) i, appearance of cells after transfer, and ii, altered cell morphology observed 30 min later. Scale bars represent 5 μm.

For a further comparison of zafrin lysis, B. subtilis cultures treated with either the cell wall inhibitor ampicillin or the protein synthesis inhibitor tetracycline were observed. After a 30-min exposure at the appropriate MIC, these cultures contained far more intact cells, hardly any debris and a large number of ‘ghost cells’ that represent intact portions of the peptidoglycan cell wall (supplementary Fig. S1). Taken alongside the absence of extensive amorphous cell debris, or intact cells with ‘visible’ septa, these results suggest that zafrin had a mode of action different from that of ampicillin or tetracycline.

Zafrin does not target the cell wall

To investigate the antibiotics' effect on cells further, it was determined whether bacterial cells treated with zafrin could be induced to form spheroplasts. In these experiments, B. subtilis was grown in an osmotically protected medium containing 0.5 M sucrose. In mid-exponential growth, aliquots of the culture were treated with various concentrations of zafrin and examined under a microscope. At concentrations between 12.5 and 100 μg mL−1, zafrin did not induce the formation of osmotically fragile spheroplasts. Instead, the majority of cells died in a manner identical to the cells grown in Luria–Bertani (LB) media. To test the hypothesis that the cell wall was not the site of zafrin action, osmotically protected B. subtilis protoplasts were exposed to the antibiotic. The protoplasts lysed, suggesting that the cell wall is not the principal target of zafrin.

Zafrin-induced lysis resembles the effect of compounds that disrupt the cell membrane

To gain further insight into the compounds mode of action, the pattern of zafrin killing was compared with that of antibiotics known to target the cell membrane. The lantibiotic nisin forms a complex with the membrane-anchored cell wall precursor molecule lipid II at low concentrations, and permeabilizes and disrupts the cell membrane at high concentrations (Breukink et al., 1999) (Hyde et al., 2006). Using B. subtilis 168 cells grown in LB broth, it was observed that at 5 μg mL−1 nisin produced ‘ghost’ cells indicative of controlled cell wall lysis (Fig. 4c), while at a high concentration nisin (50 μg mL−1) caused rapid cell lysis (Fig. 4d). Under the microscope, it was clear that the rapid lysis generated the same amorphous cell debris observed with zafrin-treated cultures (Fig. 4a i and 4b). Because this phenotype was not observed with ampicillin, vancomycin, tetracycline, kanamycin, spectinomycin or erythromycin (data not shown), it was decided to investigate another surface-active compound that has been reported to induce the rapid lysis of B. subtilis, the detergent Triton X-100 (Tilby, 1978). The addition of Triton X-100 (0.09%) to a mid-exponential culture of B. subtilis caused a rapid decline in OD, and once again produced the amorphous cell debris associated with zafrin and nisin (50 μg mL−1) treatment. The effect of Triton X-100 and zafrin on B. subtilis cells growing on agarose covered microscope slides was then analysed. At a concentration of half the MIC, both compounds produced phase-bright regions within the cytoplasm of c. 20% of the cells (Fig. 4e ii and f ii). These cells did not go on to grow and divide, and in the case of zafrin appeared to shrink by up to 20%. The size of the phase-bright regions, their distribution within the cell and the speed with which they formed in rich media meant they were not linked to spore formation. Therefore, it is concluded that the rapid lysis of zafrin, the production of extensive amorphous cell debris and the appearance of zafrin-treated cells under the microscope, are consistent with the effect produced by compounds that disrupt the cell membrane.


This paper describes the isolation, purification and elucidation of the chemical structure of a novel antibacterial compound produced by an isolate of the marine bacterium Pseudomonas stutzeri (CMG1030). Antibacterial activity among marine bacteria is a well-known phenomenon and has been reported in a number of studies (Rosenfeld & Zobell, 1947; Dopazo et al., 1988) (McCarthy et al., 1994). However, this is the first time that Pseudomonas stutzeri has been shown to produce an antibiotic. Considering the wide geographical and environmental distribution of this species and the existence of numerous genomovars with a variety of metabolic peculiarities (Lalucat et al., 2006), the isolation of CMG1030 is not particularly surprising. Indeed, a second antibacterial compound produced by a second strain of Pseudomonas stutzeri is currently being characterized in the authors' laboratory (unpublished data).

The antibacterial compound was determined to be 4β-methyl-5, 6, 7, 8 tetrahydro-1 (4ß-H)-phenanthrenone and named zafrin. To the best of the authors' knowledge, zafrin has not been described previously and does not share significant structural similarity with any known antibacterial agent. Other novel antimicrobial compounds, including polybrominated compounds, have been isolated recently from marine organisms (Handayani et al., 1997). These include an anti-MRSA compound 2, 4-diacetylphloroglucinol (DAPG) produced by a Pseudomonas sp. associated with marine alga (Isnansetyo et al., 2001) and phenolic anti-MRSA substances from Pseudoalteromonas phenolica (Isnansetyo & Kamei, 2003). Therefore, it appears that marine microorganisms will become an increasingly valuable source of novel antibacterial compounds.

The preliminary characterization of zafrin established that the compound was stable and active against a number of important clinical and environmental microorganisms. Zafrin also appeared to have a broad spectrum, exhibiting activity against a range of both Gram-positive and Gram-negative bacteria. The MIC that was recorded for zafrin (50–125 μg mL−1) compares favorably with other novel antimicrobials such as DAPG (0.5–1 mg mL−1) (Isnansetyo & Kamei, 2003). The experiments on B. subtilis showed that the killing rate of zafrin was faster than ampicillin, vancomycin or tetracycline. Therefore, it would appear that zafrin is a very potent antibacterial agent and worthy of further investigation.

Finally, the preliminary experiments to determine zafrin's likely mode of action on B. subtilis showed that zafrin does not target the cell wall and that its pattern of lysis resembles that of compounds such as nisin (50 μg mL−1) and Triton X-100 which disrupt the cell membrane. Because the chemical structure of zafrin indicates it is likely to be hydrophobic and lipophilic, it is proposed that zafrin's mode of action is via the disruption of the cytoplasmic membrane. It is also noted that the failure of zafrin to lyse Candida albicans, Schizosaccharomyces pombe or, most importantly, Pseudomonas aeruginosa suggests that zafrin has a specific mechanism for disrupting the cell membrane.


The authors wish to express their sincere gratitude to the Higher Education Commission for a merit Scholarship to one of the authors Bushra Uzair; the authors are also grateful to Dr B. Siddiqu Director Spectrum Fisheries for providing samples for the isolation of bacterial strains. Thanks are due to the HEC-BC project for providing financial support for the UK trip to study the antibacterial activities of the isolated compound.