A Novel Extracellular Cyclic Lipopeptide Which Promotes Flagellum ...

JOURNAL OF BACTERIOLOGY, Mar. 1992, p. 1769-1776

Vol. 174, No. 6

0021-9193/92/061769-08$02.00/0

Copyright X 1992, American Society for Microbiology

A Novel Extracellular Cyclic Lipopeptide Which Promotes Flagellum-Dependent and -Independent Spreading Growth of Serratia marcescens TOHEY

MATSUYAMA,'*

KENJI KANEDA,1 YOJI NAKAGAWA,2 KIMIO ISA,3 HISAKO HARA-HOTTA,4 AND IKUYA YANO4

Department of Bacteriology, Niigata University School of Medicine, Niigata 951,1 Division of Chemistry, General Education Department, Niigata University, Niigata 950,2 Department of Chemistry, Faculty of Education, Fukui University, Fukui 910, and Department of Bacteriology, Osaka City University Medical School, Osaka 545,4 Japan Received 24 September 1991/Accepted 8 January 1992

Serrawettin W2, a surface-active exolipid produced by nonpigmented Serratia marcescens NS 25, was examined for its chemical structure and physiological functions. The chemical structure was determined by degradation analyses, infrared spectroscopy, mass spectrometry, and proton magnetic resonance spectroscopy. Serrawettin W2 was shown to be a novel cyclodepsipeptide containing a fatty acid (3-hydroxydecanoic acid) and five amino acids. The peptide was proposed to be D-leucine (N-bonded to the carboxylate of the fatty acid)-L-serine-L-threonine-D-phenylalanine-L-isoleucine (bonded to the 3-hydroxyl group). By examining the effects of isolated serrawettin W2 on serrawettinless mutants, this lipopeptide was shown to be active in the promotion of flagellum-independent spreading growth of the bacteria on a hard agar surface. The parent strain NS 25 formed a giant colony with a self-similar characteristic after incubation for a relatively long time (1 to 2 weeks), similar to other fractal colony-producing strains of S. marcescens (producers of the different serrawettins Wl and W3). On a semisolid medium that permitted flagellum-dependent spreading growth, an external supply of serrawettin W2 accelerated surface translocation of a serrawettinless mutant during a short period (12 h) of observation. In contrast, bacterial translocation in the subsurface space of the semisolid agar was not enhanced by serrawettins. Thus, the extracellular lipids seem to contribute specifically to the surface translocation of the bacteria by exhibiting surfactant activity.

Previously, we reported that Serratia marcescens and Serratia rubidaea demonstrate wetting activity on various hydrophilic and hydrophobic surfaces (1, 12, 13, 17). The extracellular products responsible for such activity were isolated and named serrawettins (lipopeptides produced by S. marcescens) (15, 16) and rubiwettins [,-D-glucopyranosyl 3-(3'-hydroxytetradecanoyloxy)decanoate and its glucosefree relatives produced by S. rubidaea] (13). Serrawettin Wi, produced by many pigmented S. marcescens strains, is a surface-active cyclodepsipeptide identical to serratamolide, which was discovered by Wasserman as an antibiotic (25, 26). Serrawettins W2 and W3 are also surface-active exolipids of nonpigmented strains of S. marcescens (W2 from strain NS 25 and W3 from strains NS 45 and NS 50). Although serrawettins W2 and W3 were identified as novel types of ninhydrin-negative lipopeptides giving protonated molecular ions at 732 and 684 mlz, respectively, in positive secondary ion mass spectra, the precise chemical structures remained unsolved. In contrast to serrawettin Wl, which has a symmetric dilactone structure composed of 2 mol of serine and 2 mol of 3-hydroxydecanoic acids (12, 26), W2 and W3 were shown to have asymmetric cyclic structures composed of several amino acids and fatty acids (14). In the present study, the chemical structure of serrawettin W2 was determined by a combination of conventional degradation analyses and nondegradation analyses such as two-dimensional nuclear magnetic resonance (NMR) spectroscopy. The isolated serrawettins Wl, W2, and W3 lowered the

*

surface tension of saline to 32.2, 33.9, and 28.8 mN/m, respectively (15). The extracellular amounts of these biosurfactants are also noteworthy (15 to 17% of dry weight bacterial mass) (12). The physiological roles of such surfaceactive exolipids, however, are mostly unknown. Therefore, an attempt was made to obtain mutants unable to produce serrawettins by developing a simple method called direct colony thin-layer chromatography (17) for screening colonies of mutants defective in production of the specific lipid. Culture suspensions of mutants obtained by this method lost their wetting activity on glass and polystyrene surfaces (1, 17) and failed to sink into water-repelling fibrous material (17). Previously, S. marcescens strains producing serrawettin Wl or W3 were reported to exert flagellum-independent spreading growth on a surface of solid agar. Since these mutants failed to form such giant colonies, serrawettins Wl and W3 were suggested as the promoters of such a novel type of spreading growth (16). In the present study, isolated serrawettin W2 was externally supplied to the mutants and examined for its role in the translocation behavior of the bacteria on solid and semisolid agar surfaces. The results indicated that serrawettins are the potent extracellular product enhancing flagellum-dependent and flagellum-independent spreading growth of the bacteria in a surface environment.

MATERLALS AND METHODS

Bacterial strains and growth. The Serratia marcescens strains used in this study are listed in Table 1. Mutants NS 25-04 and NS 25-23 were obtained by UV irradiation and

Corresponding author. 1769

1770

J. BACTERIOL.

MATSUYAMA ET AL. TABLE 1. S. marcescens strains used Strain

NS NS NS NS NS NS NS NS NS

25 (wild type) 25-04 25-23 38 (wild type) 38-09 38-45 45 (wild type) 45-09 45-11

a +, yes; -, b

Serrawettin

Presence of

Reference

produced

flagellaa

or source

W2 __b W2 Wi

+ + + + + + -

12 This study This study 12 17 16 15 16 16

Wi W3 W3

no.

-,none.

nitrosoguanidine mutagenesis (18), respectively. The screening method, direct colony thin-layer chromatography (17), was carried out to examine the production of each serrawettin. To obtain serrawettin W2, strain NS 25 was grown at 30°C for 3 days on a peptone glycerol agar medium as described previously (13). To examine the flagellum-independent spreading, the bacteria were point inoculated with a toothpick onto Vogel-Bonner minimum agar (24) containing 1.4% agar (Eiken, Tokyo, Japan). For flagellum-dependent bacterial spreading, Nutrient agar (Eiken) containing 0.5% agar was used. These test media for bacterial spreading were enriched with specified concentrations of glucose and openair dried. Incubation was carried out in a humidified environment (relative humidity, 88% 2%). Preparation of serrawettin W2. A wet bacterial mass grown on the surface of peptone glycerol agar was collected with a rubber scraper and mixed with 10 volumes of ethanol. After removal of sediment by centrifugation, ethanol was removed by evaporation, and then the dry material was further extracted with chloroform-methanol (2:1, vol/vol). The extracts were developed on a thin-layer chromatography (TLC) plate of Silica Gel G (Analtech, Inc., Newark, Del.) in solvent system I (chloroform-methanol-5 M ammonia [80:25:4, vol/vol/vol]). The separated band was scraped off and extracted with chloroform-methanol. Two other solvent systems (acidic and neutral) were used in TLC to examine the purity of the extract as described previously (13). The preparative TLC was repeated to obtain serrawettin W2, giving a single spot in TLC with three different solvent systems. To remove the remaining cations, the preparation was passed through a DEAE-cellulose SH (Serva, Germany) column. Degradation and derivatization methods. Serrawettin W2 was acid hydrolyzed in 6 M HCl at 120°C for 90 min. The resultant n-hexane-soluble products were transmethylated in benzene-methanol-concentrated H2SO4 (10:20:1, vol/vol/ vol) for gas-liquid chromatography-mass spectrometry (GLC-MS) analyses. The methanol-soluble products were dried after the repeated addition of ethanol and stored for chemical composition analyses. Alkaline hydrolysis was performed in 0.2 M sodium hydroxide-chloroform-methanol-water (1:2:0.02, vol/vol/vol) by shaking at 37°C for 2 h. After partition between water and chloroform-methanol in the presence of ion-exchange resin (IR-120P; Amberlite), chloroform-methanol phases were evaporated to dryness and examined by TLC. The product giving a single spot (W2A) in TLC was methylated by treatment with an ethereal solution of diazomethane as described previously (13). The product thus methylated was designated W2AM.

All degradation and derivatization products were checked by TLC with solvent system I and n-hexane-diethyl ether (4:1, vol/vol) for nonpolar products and chloroform-methanol-acetic acid-water (60:40:12:5 by volume) for polar products. Physicochemical analysis. GLC-MS was carried out on a Hitachi M-80B double-focusing mass spectrometer. Conditions for the gas chromatographic column, the inlet, and the recording were as described previously (13). Infrared spectra were recorded with a model DS-402G apparatus (Japan Spectroscopic Co., Tokyo, Japan) as a KBr disk. A Hitachi 835 amino acid analyzer was used for amino acid composition analysis. A C-terminal amino acid was determined by the conventional method following the reduction of W2 with lithium borohydride. The presence of D-amino acids was determined by injecting the sample into high-pressure liquid chromatography (HPLC) (Shimadzu LC-6A) equipped with a CROWNPAK CR (+) column (0.4 by 15 cm) (Daicel, Tokyo, Japan). Amino acids were eluted with 35.8 mM HCl04 (pH 1.5; kept at 1°C) at a flow rate of 0.4 ml/min. The wavelength of the detector was set at 200 nm. Secondary ion mass spectrometry (SIMS) was performed on the mass spectrometer described above by using a beam of xenon ions at an accelerating voltage of 8 (positive SIMS) or 7 (negative SIMS) kV. Glycerol was used as a matrix. Fast atom bombardment tandem mass spectrometry was carried out by using a JEOL DX 303 in a setup of EBE (E and B represent electrostatic and magnetic sectors, respectively) operating at an ion-accelerating voltage of 3 kV. Argon was used as the collision gas at a pressure sufficient to reduce the precursor ion signal by 50%. As a matrix, m-nitrobenzyl alcohol was used. Proton NMR spectra were recorded on a JEOL JNMGSX-500 spectrometer operating at 500.2 MHz. The sample (4.3 mg) was dissolved in 0.6 ml of [2HJdimethyl sulfoxide (Merck & Co., Inc., Rahway, N.J.), and spectra were obtained at 27°C. Tetramethylsilane was used as an internal standard. One-dimensional spectra were obtained with a sweep width of 8 kHz. Two-dimensional nuclear Overhauser enhancement spectroscopy (NOESY) was performed in the phase-sensitive mode with a spectral width of 5 by 5 kHz. Experiments were recorded with 0.205-s acquisition time, 1.795-s pulse delay time, and 800-ms mixing time. Doublequantum filtered two-dimensional correlation spectroscopy was performed in the phase-sensitive mode with a spectral width of 5 by S kHz. Experiments were recorded with 0.205-s acquisition time and 1.795-s pulse delay time. RESULTS of Analyses acid hydrolysates. n-Hexane-soluble products after acid hydrolysis of serrawettin W2 were transmethylated and examined by GLC-MS. Methyl esters of 3-hydroxydecanoic acid and decenoic acid were identified from specific mass spectra especially by the presence of mlz 103 ion (3-OH, C10:0) and M - 32 ion (C10:1). GLC-MS also indicated the percentage compositions of the methyl esters. Compositions of 3-hydroxydecanoic acid and decenoic acid were estimated to be 82.3 and 17.7%, respectively. Methanol-soluble products of acid hydrolysis of W2 were all ninhydrin positive. By the amino acid analyzer, these were shown to be serine, threonine, leucine, phenylalanine, isoleucine, and valine (molar ratio, 1:1:1:1:0.7:0.2). Leucine and phenylalanine had 10.75- and 28.63-min retention times,

STRUCTURE AND FUNCTION OF SERRATIA BIOSURFACTANT

VOL. 174, 1992

1771

762

(M-H) 2AN

W2AM

V2

748 634

(M-H)F

71

730

thAILIJM II . A1jadills...^L .1 Id 1.I. I

t

(N-H)

W2A

414

....

W2

578 W2A

I

68

600

a_-L v --WJMj.l... J~~~~~~~_n A Al It ---~~~. .sM A X, I 7flI -,

-

--800

360

3341 L--~A6 ],,40x6J JIIA W Lu 400

62 h.LA--A

500

-

JL-

L L

j

iLn I IAin aI

,&l__ .

-.

600

700

J. -L

m/z FIG. 2. Segments of negative secondary ion mass spectra of serrawettin W2 (W2), its alkaline hydrolysates (W2A), and the diazomethane-treated alkaline hydrolysates (W2AM). Spectra of W2A and W2AM are given as insets in the mass spectrum chart of serrawettin W2.

0 1

2

3

FIG. 1. Thin-layer chromatograms of serrawettin W2 (lane 1), its alkaline hydrolysates (W2A, lane 2), and the diazomethane-treated alkaline hydrolysates (W2AM, lane 3). Each sample was developed with solvent system I. The plate was sprayed with 50% (vol/vol) H2SO4 and heated at 200°C for 20 min. 0 denotes origin of the development.

respectively, in analyses by HPLC. Under the same elution conditions, the standard amino acids demonstrated the following retention times (in minutes): D-leucine, 10.73; L-leucine, 28.35; D-phenylalanine, 27.71; L-phenylalanine, 48.83. Thus, leucine and phenylalanine were identified as D-amino acids. Other amino acids were shown to be L-amino acids by the same analytical method (data not shown). Analyses of alkaline hydrolysates. TLC of alkaline hydrolysates of serrawettin W2 gave a ninhydrin-negative tailed spot (W2A in Fig. 1) with an increased polarity (a shift in Rf values from 0.86 to 0.42). By treatment of W2A with diazomethane, a nonpolar derivative was generated and gave a round spot (W2AM in Fig. 1). In a comparative infrared analysis, an ester-linkage absorption band (at 1,730 cm-1) in a spectrum of W2 was not present in that of W2A, and a strong absorption band of a hydroxyl group (at 3,400 cm-') became dominant in a spectrum of W2A. In negative SIMS analyses, (M H)- ions of W2, W2A, and W2AM were clearly recognized (Fig. 2). In accordance with the development of a single product (W2A) after alkaline hydrolysis of W2 and the disappearance of the ester-linkage absorption band from an infrared spectrum of W2A, the mass unit shift from m/z 730 (W2) to m/z 748 (W2A) may indicate the hydrolytic opening of an intramolecular lactone ring. A generated free carboxyl group may then contribute to the tailing characteristic of the W2A spot in TLC and may be an acceptor moiety for the methylation -

by diazomethane. A plus-14 mass unit shift from W2A to W2AM is consistent with this chemical change. The results described above seem to indicate that serrawettin W2 is a novel cyclodepsipeptide with a molecular weight of 731. Thus, W2 may be a condensation product of equimolar amounts of 3-hydroxydecanoic acid, serine, threonine, leucine, phenylalanine, and isoleucine. In positive SIMS (data not shown), a corresponding ion (M + H)+ of W2 was also clearly observed at m/z 732. Since there were no specific ions at m/z 712 in negative SIMS (Fig. 2) and at mlz 714 in positive SIMS of W2, decenoic acid in acid hydrolysates may be a dehydration product of 3-hydroxydecanoic acid. W2 containing valine instead of isoleucine seems to be present as a minor component. The ion at minus-14 mass units from the main molecular or pseudomolecular ion may correspond to this minor isomer of serrawettin W2. Tandem mass spectrometry analysis. As W2 is ninhydrin negative, 3-hydroxydecanoic acid seems to be linked to the N terminus of a pentapeptide. Therefore, it was impossible to examine the sequence of these amino acids by an ordinary method. The C-terminal amino acid, however, was identified as isoleucine by determining the missing amino acid in acid hydrolysates of W2 after reduction with lithium borohydride. In Fig. 3, a mass spectrum obtained by tandem mass spectrometry of W2AM is shown. Specific fragment ions resulting from degradation at each peptide linkage were recognizable. By referring to the mass unit of the respective amino acids, the amino acid sequence Leu-Ser-Thr-Phe-Ile seemed to be the most probable. Although tandem mass spectrometry of W2 itself also gave a mass spectrum consistent with the above proposed sequence (data not shown), only degradation ions resulting from sequential loss of the C-terminal amino acid were observed. In the mass spectrum of W2AM (Fig. 3), sequential degradation ions from both

1772

MATSUYAMA ET AL.

J. BACTERIOL.

acid was indicated for an ester linkage with a carboxyl group of isoleucine. NOE (nuclear Overhauser effect) interactions among amide protons examined by phase-sensitive NOESY are indicated by the lines connecting the peaks in Fig. 5. Appearing as a cross peak (FNH-INH), the amide proton of isoleucine is connected with the amide proton of phenylalanine. The connection is also manifested by a second cross peak in a symmetrical location with respect to the diagonal. Then, NOEs between the amide protons of phenylalanine and threonine, threonine and serine, and serine and leucine are traceable sequentially. Thus, the sequence Leu-Ser-ThrPhe-Ile suggested by tandem mass spectrometric analysis was consistent with the NMR analysis. In addition, a sequential examination of NOEs between ao proton a; and amide proton Ni,1 by the phase-sensitive NOESY indicated the same amino acid sequence (data not shown). The proposed structure of the major component of serrawettin W2 is shown in Fig. 6. Physiological function of serrawettin W2. On a hard agar plate that does not permit flagellum-dependent spreading growth of S. marcescens (16), the wild-type strain NS 25 demonstrated extended spreading growth (Fig. 7A). The colony exhibited branching with a self-similar characteristic (the fractal dimension was not determined, mainly because of fused unclear boundaries among these branches). On the other hand, mutants unable to produce W2 failed to exert such extended spreading growth (Fig. 7B). The longest diameter of each colony developed after 2 weeks of cultivation was measured. The values obtained from the wild-type (NS 25) and mutant (NS 25-04) colonies were 37.2 ± 4.9 and 15.1 ± 0.6 mm (mean + SD; n = 5), respectively. Three independent mutants defective in production of W2 have exhibited such a defective spreading phenotype (data not shown). These mutants, however, were all accompanied by the loss of unidentified glycolipids that were reported previously (17). We examined the external effect of isolated W2 on the spreading growth of mutant strains. As is shown in Fig. 8, the spreading growths of the mutants are prominent near the part of the disk containing a large amount of W2. On semisolid nutrient agar, wild-type strains of S. marcescens demonstrated marked surface spreading growth in a

R

e

b u

n

d a

cn

M/Z

FIG. 3. Tandem mass spectrum of W2AM. Proposed cleavage sites and fragments are shown by indicating the mlz values of the corresponding main peaks. Peaks at mlz 86 and 120, labeled xL and F, are indicative of Leu and/or Ile and Phe, respectively.

sides (from the acylated N terminus and from the methylesterified C terminus) were evident. Proton NMR. Since the amino acid sequencing by tandem mass spectrometry was not definitive and the location of the lactone ring (three possible hydroxyl groups are present, i.e., in 3-hydroxydecanoic acid, serine, and threonine) must be determined, proton NMR analysis of W2 was carried out. A normal spectrum of W2 is given in Fig. 4. The fine assignments of these protons were performed by referring to the results of double-quantum filtered two-dimensional correlation spectroscopy (data not shown). Two peaks of hydroxyl proton (TOH and SOH) demonstrated cross peaks with 1 and 1' protons of serine and threonine. Thus, the hydroxyl groups of these amino acids were shown to be not bonded to the carboxyl group of isoleucine. In addition, the corresponding methine and methylene protons of these amino acids demonstrated chemical-shift values in the ordinary range (3.5 to 4.5 ppm), as shown in Fig. 4. On the other hand, a methine proton (XCH) of an acyl group demonstrated a chemical shift (4.93 ppm; see Fig. 4) which is consistent with the shift due to lactone ring formation (lower field shift). Consequently, the hydroxyl group of 3-hydroxydecanoic

I 0 A-

_ Ij

j j

8

a

jJr

_

7

6

5

4

3

2

1

FIG. 4. One-dimensional proton NMR spectrum of serrawettin W2.

0


VOL. 174, 1992

1773

I1H

f

*11

FIG. 5. Phase-sensitive NOESY spectrum of serrawettin W2. The amide proton region is shown. NOE connectivities between neighboring amide protons are sequentially indicated by the solid connecting lines.

short time (18 to 24 h), but flagellumless mutants (e.g., NS 25-23) did not. Flagellated mutants (NS 38-09, NS 25-04, and NS 45-09) demonstrated surface spreading growth, but at a remarkably reduced speed. The role of serrawettin in flagellum-dependent spreading growth was examined by supplying W2 externally to mutant NS 25-04 growing on semisolid media. After cultivation of the mutant NS 25-04 for 24 h, the strain demonstrated surface spreading growth to the degree shown in Fig. 9A. However, when a paper disk containing 100 ,ug of W2 was placed near the growing colony at 12 h postinoculation, thin spreading growth covering the total agar surface was induced, as was observed at 24 h (Fig. 9B). An extracellular complementation between the mutant NS 25-04 and the mutant NS 25-23 (nonflagellated but a W2 producer) was observed with a mixed inoculation of the two mutants. The Wi- or W3-producing nonflagellated mutants (NS 38-45 and NS 45-11) also complemented the spreading growth of the mutant NS 25-04. In Fig. 10,

CH3(CH2)*CHCH2CO-D-Leu-L-Ser-L-Thr-D-Phe-L-lIe I

I

FIG. 6. Proposed structure of serrawettin W2.

complementationbetween Wi-less flagellated NS 38-09 and W3-producing nonflagellated NS 45-11 is shown. Since NS 38-09 was a pigmented strain (prodigiosin producer) and NS 45-11 was not, the presence of both mutants in the swarming bacterial mass was easily recognized by single-colony isolations. The promotive function of serrawettins on flagellumdependent spreading growth of S. marcescens was observed only with bacteria growing in a surface environment. No promotive effect of the serrawettin W2 was observed on the spreading growth of mutant NS 25-04 in the subsurface space in semisolid agar (data not shown). After stab inoculation, no differences between wild types (NS 38, NS 25, and NS 45) and mutants (NS 38-09, NS 25-04, and NS 45-09) were seen in spreading growths in semisolid agar. It is also noteworthy that the flagellum-dependent surface spreading growth of S. marcescens progressed mostly in a spiral clockwise manner (Fig. 9 and 10) irrespective of the position (upside down or vertical) of the agar surface. Among 483 colonies of strain NS 25, 477 colonies demonstrated clockwise spiraling. The remaining six colonies exhibited branching with ambiguous directions. No colonies of strain NS 25 demonstrated clear counterclock-

wise spiraling.

1774

MATSUYAMA ET AL.

J. BACTERIOL.

FIG. 8. Effect of serrawettin W2 on the spreading growths of serrawettinless mutants. Sterile paper disks soaked with 5 p.l of ethanol containing 200 (upper disk) or 50 (middle disk) ,ug of serrawettin W2 or ethanol alone (lower disk) were dried and placed between the mutant NS 25-04 point inoculated on Vogel-Bonner medium (for details, see the legend to Fig. 7). The plate was incubated at 30°C for 10 days.

FIG. 7. A colony of S. marcescens on hard agar. (A) Strain NS 25 (wild type); (B) serrawettinless mutant NS 25-04. The bacteria were point inoculated onto the center of Vogel-Bonner medium containing 1.4% agar and 0.4% glucose (7 ml in a 90-mm-diameter dish) and incubated at 30°C for 2 weeks.

DISCUSSION In the present study, serrawettin W2 was shown to be a novel cyclodepsipeptide with the proposed structure 3-hydroxyldecanoyl-D-leucyl-L-seryl-L-threonyl-D-phenylalanylL-isoleucyl lactone. Thus, hydroxyl groups of serine and threonine may be hydrophilic parts of this C10 acylated pentapeptide. This is similar to serrawettin Wl possessing two hydroxyl groups of serines as hydrophilic parts. Although the precise chemical structure of serrawettin W3 has not yet been determined, degradation analyses indicated the presence of a dodecanoic acid and five amino acids (threonine, serine, valine, leucine, and isoleucine; 1:2:1:0.5:0.5 by molar ratio) (14). Therefore, serrawettins Wl, W2, and W3 have no amino acid residues with ionic hydrophilicity. On the other hand, surfactin, a famous biosurfactant produced by Bacillus subtilis, possesses the chemical structure 3-hydroxy-13-methyltetradecanoyl-L-glutamyl-L-leucyl-D-leucylL-valyl-L-aspartyl-D-leucyl-L-leucyl lactone (10). Carboxyl groups of Glu and Asp residues are hydrophilic parts of the surfactin molecule. Therefore, serrawettins are similar to surfactin with respect to the cyclodepsipeptide structure but dissimilar with respect to hydrophilic parts. Surfactin is an

anionic surfactant, whereas serrawettins are nonionic surfactants. Surfactants of microbial origin are considered superior to synthetic surfactants because of their biodegradable and potentially less toxic properties and have been examined for their usefulness in human daily life and in industrial applications (4, 23). However, we are interested in such surfaceactive products from the viewpoint of basic bacteriology. Most of these compounds (e.g., lipopeptide-containing D-amino acids) have complex structures synthesized by nonribosomal enzymatic processes under genetic controls (20, 21). Some functions beneficial to the bacteria may be expressed by the extracellular products (23). Rhamnolipids produced by Pseudomonas aeruginosa or sophorolipids produced by Torulopsis bombicola are known to function as emulsifiers for the uptake of hydrocarbons by these microbes (6, 7, 9, 11). Production of these glycolipid biosurfactants is induced in the presence of the hydrocarbons (4, 8). On the other hand, serrawettins were produced by S. marcescens in the absence of such hydrocarbons. Production of surfactin by B. subtilis is inhibited by the addition of hydrocarbons to a medium (3). Serrawettins seem, then, to have other physiological roles. Since some cyclic lipopeptides have antibiotic activities, serrawettin W2 was examined for this type of activity. MICs of serrawettin W2 against Escherichia coli ATCC 25922, P. aeruginosa ATCC 27853, B. subtilis ATCC 6633, and Staphylococcus aureus ATCC 25923 were >100, >100, 25, and 25 ,ug/ml, respectively (19). Therefore, the antimicrobial activity of serrawettin W2 was shown to be weak against the above representative strains and seemed not to be a principal function of the product. The antimicrobial activity of serrawettin Wl (serratamolide) is also reported to be not so strong (26), and studies on this antibiotic ceased long ago. S. marcescens is a ubiquitous organism inhabiting water,

VOL. 174, 1992


1775

FIG. 10. Extracellular complementation of the spreading growth by flagellumless and serrawettinless mutants on a semisolid medium. The left point was inoculated with S. marcescens NS 45-11 (flagellumless; producer of serrawettin W3), the right point was inoculated with NS 38-09 (flagellated; serrawettinless), and the center point was inoculated with a mixture of both strains. The nutrient agar contained 0.5% agar, and incubation was carried out at 30°C for 24 h.

FIG. 9. Effect of serrawettin W2 on the spreading growth of a serrawettinless mutant on semisolid media. Mutant NS 25-04 was point inoculated on nutrient agar containing 0.5% agar and 0.4% glucose and then incubated at 30°C for 24 h. Paper disks containing no (A) or 100 p.g of (B) serrawettin W2 were placed near the growing colony at 12 h postinoculation.

soil, plants, insects, and vertebrates. In such habitats, actual colonization sites were supposed to be the surfaces with various characteristics (hydrophobic, hydrophilic, uneven, fractal, smooth, axenic, living, etc.). To reach such surfaces efficiently, bacteria may have special strategies. By comparing serrawettin-producing bacteria and nonproducing bacteria, we previously showed that serrawettins enable the bacterial suspension to sink into a water-repelling fibrous material (cotton) (18). Without serrawettins, the bacterial suspension remained a droplet and the bacteria failed to reach the surfaces of cotton fibers located inside. Rubiwettins produced by S. rubidaea also had the same functions (13). Although the surface of a plant leaf is covered with a

hydrophobic cuticular wax layer, it is a known habitat for microbes (5). Bunster et al. suggested the importance of bacterial wetting activity in such phyllospheres. A suspension of Pseudomonas spp. was examined for wetting activity with special reference to the production of surface-active materials (2). As shown in the present investigation, bacterial behavior on a hydrophilic surface (surfaces of solid and semisolid agar) was also profoundly influenced by the bacterial extracellular surfactant. Since the bacteria are so small, the effects of intermolecular forces are strong enough to govern their movements. By excreting surface-active products such as serrawettins, bacteria seem to lighten the surface-tension burden working at the colony periphery. Extended spreading growths of S. marcescens shown in Fig. 7 through 10 revealed the roles of serrawettins in such a surface environment. In contrast to proteins, lipids tend to be the forgotten components of microbes (22). Microbial lipids have rarely been examined for their physiological functions. Most studies on lipids are done with structural lipids (membrane lipids). Herein, serrawettins are shown to be a novel type of exolipid with a specific function. More studies on the function of microbial lipids will provide new insight into understanding bacterial behaviors in nature. ACKNOWLEDGMENTS We thank S. Odani for amino acid analysis, K. Shizukuishi for negative SIMS analysis, T. Imanari for NMR analysis, and C. G. Woodward for help in the preparation of the manuscript. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of

Japan (no. 02808026).

1776

MATSUYAMA ET AL.

REFERENCES 1. Bar-Ness, R., N. Avrahamy, T. Matsuyama, and M. Rosenberg. 1988. Increased cell surface hydrophobicity of a Serratia marcescens NS 38 mutant lacking wetting activity. J. Bacteriol. 170:4361-4364. 2. Bunster, L., N. J. Fokkema, and B. Schippers. 1989. Effect of surface-active Pseudomonas spp. on leaf wettability. Appl. Environ. Microbiol. 55:1340-1345. 3. Cooper, D. G., C. R. Macdonald, S. J. B. Duff, and N. Kosaric. 1981. Enhanced production of surfactin from Bacillus subtilis by continuous product removal and metal cation additions. Appl. Environ. Microbiol. 42:408-412. 4. Cooper, D. G., and J. E. Zajic. 1980. Surface-active compounds from microorganisms. Adv. Appl. Microbiol. 26:229-253. 5. Fokkema, N. J. 1988. Agrochemicals and beneficial role of phyllosphere yeasts in disease control. Ecol. Bull. 39:91-93. 6. Hisatsuka, K. I., T. Nakahara, N. Sano, and K. Yamada. 1971. Formation of rhamnolipid by Pseudomonas aeruginosa and its function in hydrocarbon fermentation. Agric. Biol. Chem. 35: 686-692. 7. Ito, S., and S. Inoue. 1982. Sophorolipids from Torulopsis bombicola: possible relation to alkane uptake. Appl. Environ. Microbiol. 43:1278-1283. 8. Itoh, S., H. Honda, F. Tomita, and T. Suzuki. 1971. Rhamnolipids produced by Pseudomonas aeruginosa grown on n-paraffin. J. Antibiot. 24:855-859. 9. Itoh, S., and T. Suzuki. 1972. Effect of rhamnolipids on growth of Pseudomonas aeruginosa mutant defective in n-paraffinutilizing ability. Agric. Biol. Chem. 36:2233-2235. 10. Kakinuma, A., M. Hori, H. Sugino, I. Yoshida, M. Isono, G. Tamura, and K. Arima. 1969. Determination of the location of lactone ring in surfactin. Agric. Biol. Chem. 33:1523-1524. 11. Koch, A. K., 0. Kippeli, A. Fiechter, and J. Reiser. 1991. Hydrocarbon assimilation and biosurfactant production in Pseudomonas aeruginosa mutants. J. Bacteriol. 173:4212-4219. 12. Matsuyama, T., M. Fujita, and I. Yano. 1985. Wetting agent produced by Serratia marcescens. FEMS Microbiol. Lett. 28:125-129. 13. Matsuyama, T., K. Kaneda, I. Ishizuka, T. Toida, and I. Yano. 1990. Surface-active novel glycolipid and linked 3-hydroxy fatty

J. BACTERIOL.

14.

15. 16. 17.

18.

19. 20.

21.

22. 23. 24. 25.

26.

acids produced by Serratia rubidaea. J. Bacteriol. 172:30153022. Matsuyama, T., K. Kaneda, and I. Yano. 1986. Two kinds of bacterial wetting agents: aminolipid and glycolipid. Proc. Jpn. Soc. Mass Spectrom. 11:125-128. (In Japanese.) Matsuyama, T., T. Murakami, M. Fujita, S. Fujita, and I. Yano. 1986. Extracellular vesicle formation and bio-surfactant production by Serratia marcescens. J. Gen. Microbiol. 132:865-875. Matsuyama, T., M. Sogawa, and Y. Nakagawa. 1989. Fractal spreading growth of Serratia marcescens which produces surface active exolipids. FEMS Microbiol. Lett. 61:243-246. Matsuyama, T., M. Sogawa, and I. Yano. 1987. Direct colony thin-layer chromatography and rapid characterization of Serratia marcescens mutants defective in production of wetting agents. Appl. Environ. Microbiol. 53:1186-1188. Matsuyama, T., and H. Uetake. 1972. Chromosomal locations of Salmonella conversion phages: mapping of prophages g341, -15 F3 in Salmonella anatum. Virology 49:359-367. Nakagawa, Y., and T. Matsuyama. Unpublished data. Nakano, M. N., M. A. Marahiel, and P. Zuber. 1988. Identification of a genetic locus required for biosynthesis of the lipopeptide antibiotic surfactin in Bacillus subtilis. J. Bacteriol. 170:5662-5668. Nakano, M. N., and P. Zuber. 1989. Cloning and characterization of srJB, a regulatory gene involved in surfactin production and competence in Bacillus subtilis. J. Bacteriol. 171:53475353. Ratledge, C., and S. G. Wilkinson. 1988. Preface, p. xvii-xviii. In C. Ratledge and S. G. Wilkinson (ed.), Microbial lipids. Academic Press, London. Rosenberg, E. 1986. Microbial surfactants. Crit. Rev. Biotech. 3:109-132. Vogel, H. G., and D. M. Bonner. 1956. Acetylornithinase of Eschenchia coli: partial purification and some properties. J. Biol. Chem. 218:97-106. Wasserman, H. H., J. J. Keggi, and J. E. McKeon. 1961. Serratamolide, a metabolic product of Serratia. J. Am. Chem. Soc. 83:4107-4108. Wasserman, H. H., J. J. Keggi, and J. E. McKeon. 1961. The structure of serratamolide. J. Am. Chem. Soc. 84:2978-2982.