Vol. 60, No. 1APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Jan. 1994, p. 31-380099-2240/94/$04.00+0Copyright X 1994, American Society for Microbiology

Structural and Immunological Characterization of a

Biosurfactant Produced by Bacillus licheniformis JF-2SUNG-CHYR LIN,' MARK A. MINTON,2 MUKUL M. SHARMA,3 AND GEORGE GEORGIOUl*

Departments of Chemical Engineering,' Chemistry,2 and Petroleum Engineering,3The University of Texas at Austin, Austin, Te-xas 78712

Received 30 August 1993/Accepted 28 October 1993

BaciUlus licheniformis JF-2 produces a very active biosurfactant under both aerobic and anaerobicconditions. We purified the surface-active compound to homogeneity by reverse-phase C18 high-performanceliquid chromatography and showed that it is a lipopeptide with a molecular weight of 1,035. Amino acidanalysis, fast atom mass and infrared spectroscopy, and, finally, 'H, '3C, and two-dimensional nuclearmagnetic resonance demonstrated that the biosurfactant consists of a heterogeneous Cl5 fatty acid tail linkedto a peptide moiety very similar to that of surfactin, a lipopeptide produced by BaciUus subtilis. Polyclonalantibodies were raised against surfactin and shown to exhibit identical reactivity towards purified JF-2lipopeptide in competition enzyme-linked immunosorbent assays, thus providing further evidence for thestructural similarity of these two compounds. Under optimal conditions, the B. licheniformis JF-2 biosurfactantexhibits a critical micelle concentration of 10 mg/liter and reduces the interfacial teinsion against decane to 6xlo-3 dyne/cm, which is one of the lowest interfacial tensions ever reported for a microbial surfactant.

Microbial compounds which exhibit pronounced surfaceactivity are classified as biosurfactants. Microbial biosurfac-tants include a wide variety of surface- and interfaciallyactive compounds, such as glycolipids, lipopeptides, po-lysaccharide-protein complexes, phospholipids, fatty acids,and neutral lipids (6). Biosurfactants consist of distincthydrophilic and hydrophobic moieties. The former can beeither ionic or nonionic and consist of mono-, di-, or polysac-charides; carboxylic acids; amino acids; or peptides. Thehydrophobic moieties are usually saturated, unsaturated, orhydroxylated fatty acids. Biosurfactants are easily biode-gradable and thus are particularly suited for environmentalapplications such as bioremediation and the dispersion of oilspills (8, 20-22).Among the many classes of biosurfactants, lipopeptides

are particularly interesting because of their high surfaceactivities and therapeutic potential. For example, surfactin,a well-studied lipopeptide antibiotic produced by Bacillussubtilis, is not only a very effective biosurfactant (5) but isalso an inhibitor of fibrin clotting (1, 3) and cyclic AMPphosphodiesterase (10).

Bacillus lichenifornis JF-2, isolated from oil-field injectionbrine (13), has been shown to be able to grow and produce avery effective biosurfactant under both aerobic and anaero-bic conditions at a very wide range of temperatures and inthe presence of high concentrations of salts (12, 17). Wehave studied the effects of various environmental parameterson the production of the biosurfactant, the formation offermentation end products, and the growth of B. licheni-formis JF-2 in batch cultures (17, 18). Although it has beenspeculated that the JF-2 biosurfactant is similar to surfactinfrom B. subtilis (7, 19), the chemical structure of the JF-2biosurfactant had not been characterized. Recently, twoother Bacillus isolates, B. licheniformis 86 (9) and another B.licheniformis strain isolated by Jenny et al. (14), have been

* Corresponding author. Mailing address: Department of Chemi-cal Engineering, The University of Texas at Austin, Austin, Texas78712. Phone: (512) 471-6975. Fax: (512) 471-7963. Electronic mailaddress: gg@mail.che.utexas.edu.

shown to produce lipopeptides with peptide moieties con-taining C-terminal amino acid residues different from thoseof surfactin.

In this study, we purified the surface-active compoundfrom B. licheniformis JF-2 to apparent homogeneity. Thestructure of this compound was characterized by amino acidanalysis and various spectroscopic techniques. As part ofthe characterization studies, we raised polyclonal antibodiesagainst surfactin and showed that they exhibit the samereactivity for the JF-2 surfactant. This result together withamino acid analysis and two-dimensional nuclear magneticresonance (NMR) data indicated that the peptide sequenceof the JF-2 surfactant is identical to that of surfactin.However, the two compounds differ with respect to thecomposition of their fatty acid tails. In addition to thechemical characterization of the JF-2 lipopeptide, we studiedits interfacial properties and found that under optimal con-ditions the JF-2 lipopeptide is one of the most effectivebacterial lipopeptide surfactants known.

MATERUILS AND METHODS

Microorganisms and growth conditions. B. lichenifonnisJF-2 (ATCC 3097) was obtained from the American TypeCulture Collection (Rockville, Md.). The bacteria weregrown aerobically in a mineral salt medium (13), containing0.1% (NH4)2SO4, 0.025% MgSO4, 1% (wt/vol) glucose, 0.5%NaCl in 100 mM phosphate buffer (pH 7.0), and 1.0%(vol/vol) trace metals solution, in 2-liter Erlenmeyer flaskswith a working volume of 1 liter at 42°C for 15 h. The tracemetals solution contained 0.1% (wt/vol) EDTA, 0.3%MnSO4, 0.001% FeSO4, 0.01% CaCl2, 0.01% CoCl2, 0.01%ZnS04, 0.001% CuS04, 0.001% AlK(S04)2, 0.001% H3B04,and 0.001% Na2MoO4 (4).

Isolation and purification. The cells were removed fromthe culture by centrifugation at 8,000 x g for 15 min.Surface-active compounds were then isolated from the clearbroth either by acid precipitation with concentrated HCl atpH 2.0 (7) or by XAD-2 (Sigma, St. Louis, Mo.) adsorptionchromatography. For the former, the surfactant-containing

31

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

APPL. ENvIRON. MICROBIOL.

precipitate was collected by centrifugation and resuspendedin 15 ml of water adjusted to pH 6.0 and subsequentlylyophilized. The lyophilized material was then extractedwith 5 ml of a mixture of chloroform and methanol (1:2[vol/vol]). For XAD-2 adsorption chromatography, the fer-mentation broth supernatant was loaded onto a column (16by 500 mm) at a flow rate of 1 ml/min. The column was elutedwith 1.5 bed volume of methanol following a 1.5 bed volumeof water wash. Nonvolatile material in the eluent wasconcentrated by evaporation in a rotary evaporator at 45°C.Samples obtained by either procedure were designated asthe crude biosurfactant preparation. The crude biosurfactantpreparation was dissolved in 3 ml of a mixture of chloroformand methanol (1:1) and further separated into five fractionsby liquid chromatography on a silica gel (no. 62; Mallinck-rodt, Paris, Ky.) column (28 by 500 mm) eluted with 5%methanol in chloroform at a flow rate of 2 ml/min. Thefraction containing the surface-active component was iden-tified by interfacial tension measurements, as describedbelow. The active compound was purified to homogeneity bypreparative reverse-phase liquid chromatography at roomtemperature with a Waters high-performance liquid chroma-tography (HPLC) system (Milford, Mass.) equipped with aWaters C18 ,uBondapak column (19 by 300 mm). The solventsystem consisted of mobile phase A (10 mM KH2PO4 bufferat pH 6.0) and mobile phase B (20% tetrahydrofuran inacetonitrile [HPLC grade; Fisher Scientific, Fair Lawn,N.J.]). The column was developed with 53% B isocraticallyat a flow rate of 2 ml/min. Biosurfactant-containing fractionswere lyophilized and extracted with methanol to removesalt. For analytical reverse-phase C18 HPLC analysis, apBondapak C18 column (7.8 by 300 mm) was used at a flowrate of 0.5 ml/min. The A210 of the eluent was monitored.

Characterization. Infrared spectroscopy was performed ona Nicolet 60SXR FT-IR spectrometer. The spectrum wasmeasured in a sample compartment purged for at least halfan hour with dry nitrogen before acquiring data, which weremeasured at a resolution of 4 cm-' and were averaged over500 scans. Baselines were electronically adjusted to zeroabsorbance for the measurement of spectral intensities.For amino acid analysis, the purified biosurfactant was

hydrolyzed in 6 M HCl at 105°C for 24 h. The hydrolysatewas subjected to an Applied Biosystems 420H Derivatizer/Analyzer with on-line 130A Separation System and 920AData Analysis Module (Foster City, Calif.). Norleucine wasadded in the samples to a final concentration of 500 pM as aninternal standard.

Fast atom bombardment-mass spectroscopy (FAB-MS)analysis was performed on a Finnigan TSQ 70 mass spec-trometer with an NBA matrix. Mass spectra were collectedfrom 100 to 1,200 AMU. Positive ions were detected.NMR analysis (1H, '3C, COSY, TOCSY, and ROESY

NMR) of JF-2 biosurfactant was performed in a Brukerhigh-field (11.9-T) NMR spectroscope, with 1,2-dideuterio-tetrachloroethane (CDCl2CDCl2) (Norell, Landisville, N.J.)as solvent at 348 K.

Production of polyclonal antibodies against surfactin. Atotal of 2 mg of surfactin (Calbiochem, San Diego, Calif.),dissolved in a 400-,ul mixture of phosphate-buffered saline(PBS) and dimethylformamide (PBS-dimethylformamide[3:1]), was mixed with 300 RI of Pierce Imject keyhole limpethemocyanin (KLH)-100 pl of 1 M [1-ethyl-3-(dimethylami-no)propyl]-carbodiimide-50 ,ul of 0.1 M N-hydroxyl-sulfo-succinimide (Pierce, Rockford, Ill.) at room temperature.The conjugation reaction was allowed to proceed for 20 minand subsequently quenched by the addition of 3.0 ml of 50

o

1-

0

Q)

I

4i9c

CZ

C4

AB

c

10 20Retention Time

(minute)

I I

30 40

FIG. 1. Analytical reverse-phase C18 HPLC chromatogram ofsurface-active material obtained from silica gel liquid chromatogra-phy. Fraction C was identified as the most active component byinterfacial tension measurements.

mM glycine (11). The conjugate was dialyzed against PBSovernight and used to immunize rabbits by subcutaneousinjection with incomplete Freund's adjuvant (Sigma) on thebasis of a standard schedule (2). Blood samples were col-lected from the ear vein 10 weeks after immunization (2weeks after the booster injection). Sera were prepared bycentrifuging the blood samples at 5,000 x g for 10 min.ELISA. Microtiter plates were coated by an overnight

incubation at 37°C with a surfactin-ovalbumin conjugatewhich was prepared in the same procedure as the surfactin-KLH conjugate used for immunization. The plate was thenblocked with 3% bovine serum albumin (BSA) in PBS byincubation at 37°C for 3 h. For competition enzyme-linkedimmunosorbent assays (ELISAs), 50 ,ul of diluted rabbitsurfactin-specific serum (1:1,000 in 3% BSA in PBS) and 100pA of sample were preincubated at 37°C for 2 h. Subse-quently, the mixture was transferred to microtiter platesprecoated with the surfactin-ovalbumin conjugate and incu-

32 LIN ET AL.

J1 1

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

B. LICHENIFORMIS JF-2 SURFACTANT 33

96.6

uJOz§60 V l'0-zcnm 50-

40-

27.5 ._ _ _4000 3500 30004000.9 WAVEM

FIG. 2. The infrared spect

bated for an additional 2 h. A 3% BSA solution in PBS andsurfactin at a concentration of 25 mg/liter also in PBSwere used as negative and positive controls, respectively.Plates were washed with deionized water 10 times; washingwas followed by the addition of 100 ,ul of diluted goatanti-rabbit immunoglobulin G-horseradish peroxidase conju-gate diluted 1:1,000 in 3% BSA in PBS (Bio-Rad, Richmond,Calif.). The plates were incubated for 2 h at 37°C and washedagain as described above. A total of 100 ,ul of substrate,1-Step ABTS (Pierce), was then added to each well. The

IMBERStrum of the JF-2 biosurfactant.

enzyme reaction was allowed to proceed for 10 min and thenstopped by the addition of 50 ,ul of 1% sodium dodecylsulfate (SDS). The A405 of the solution in each well wasmeasured with a Dynatech MR300 microtiter plate reader(Chantilly, Va.).To quantify the binding affinities of the serum towards

surfactin and the JF-2 biosurfactant, the percent inhibition ofantibody-conjugate binding by free antigens is defined asfollows (23): percent inhibition = {1 - [(AEX - AN )I(Ap0-ANeg)]} x 100, where AEXP is the A405 of the sampfe, APos

154100 -

80

60

40

20

0

1036

200

460 613329 1

600400

685 1018

800 1000

- 1.2

1.0

0.8

0.6

- 0.4

0.2

ti - 0.01200

FIG. 3. The FAB-MS spectrum of the JF-2 biosurfactant.

VOL. 60, 1994

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

APPL. ENVIRON. MICROBIOL.

6

A

3 25 4 I0

FIG. 4. Proton NMR spectrum of the JF-2 biosurfactant.

is the A405 of the positive control (colorless), and ANeg is theA405 of the negative control (green).

Interfacial properties. Interfacial tension measurementsagainst decane were performed in a spinning drop interfacialtensiometer (Model 300; University of Texas, Austin). Thecritical micelle concentration (CMC) was determined bymeasuring the interfacial tension of the biosurfactant solu-tion following serial dilution. All interfacial tensions weremeasured against decane.

RESULTS

Isolation and purification. Interfacial tension measure-ments indicated that the biosurfactant can be effectivelyisolated from the cell-free culture supernatant by either acidprecipitation or XAD-2 adsorption chromatography. Theinterfacial tension increased from 0.085 dyne/cm for thecell-free culture to more than 25 dyne/cm for the reneutral-ized acid precipitation supernatant of the culture. When acidprecipitation was used as the first purification step, 250 mg ofacid precipitate was obtained per liter of culture after lyoph-ilization. Subsequently, 110 mg of water-soluble materialremained after organic extraction. The crude biosurfactantpreparation was further separated into five fractions by silica

gel chromatography. Only one fraction, which was elutedafter 1.8 bed volumes, contained surface active material.Analytical reverse-phase C18 HPLC analysis showed thepresence of three major peaks in this material (Fig. 1).Preparative reverse-phase C18 HPLC was used to isolatethree fractions corresponding to the material in each of thethree peaks. Interfacial tension measurements indicated thatfraction C contained the most surface active compound. Ata concentration of 50 mg/liter in fresh medium with 5% NaCland adjusted to pH 6.0, the material obtained from fractionsA, B, and C gave interfacial tensions against decane of 1.317,1.873, and 0.060 mN/cm, respectively. Material from frac-tion C was extracted with an equal volume mixture ofchloroform and methanol to remove the salt from the HPLCmobile phase. Approximately 25 mg of the highly activecompound per liter of culture was obtained. Since theconcentration of the surfactant in the fermentation broth is34 mg/liter, the overall yield is approximately 70%. Thepurified material gave a single peak in analytical reverse-phase C18 HPLC (data not shown). Lack of anomalous peaksin proton and carbon NMRs, as discussed below, providesfurther confirmation of the purity of the biosurfactant. Thesame yield and final purity were obtained with either acidprecipitation or XAD-2 chromatography as the first separa-

I..

PPM 7

34 LIN ET AL.

0

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

B. LICHENIFORMIS JF-2 SURFACTANT 35

GluLeu Fatty

Fatty Acid Asp Leu Asp Acid

A1_

I

Ii

v

I,lV

Glu-Leu

Leu-AspV

IILeu-Leu

Fatty Acid-Glu

I

Leu-Asp

l

I

. . . . . . .I . . . .0.

I . . . .

.5I I

. . .

ppM 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0

FIG. 5. The ROESY spectrum of the JF-2 biosurfactant obtainedproton-proton relationships are indicated by filled arrowheads and labe

tion step. XAD-2 chromatography is likely to be bettersuited for the continuous removal of the surfactant from thefermentation broth and for scale-up purposes.

Chemical structure of the surfactant. Figure 2 shows theinfrared spectrum of the JF-2 biosurfactant. Bands charac-teristic of peptides (wave number 3,430:NH, wave number1,655:CO, and wave number 1,534:CN) and aliphatic chains(wave number 3,000 to 2,800, CH2 and CH3) were observed,indicating that this compound is a lipopeptide. Also ob-served was a band corresponding to an ester carbonyl group(wave number 1,730:CO) (17). Amino acid analysis indicatedthe presence of four different amino acid residues in thepeptide moiety of the biosurfactant. The composition wasdetermined to be glutamic acid:aspartic acid:valine:leucine= 1:1:1:4. The FAB-MS spectrum (Fig. 3) indicated that thebiosurfactant has a molecular mass of 1,035 Da.

Proton NMR (Fig. 4) showed seven NH signals (8 7.0 to7.7) and seven corresponding CH signals (8 3.9 to 4.9) for thea-amino acids of the peptide. These were readily correlatedwith one another as well as with the signals of the corre-

sponding alkyl residues via two-dimensional COSY andTOCSY spectra (data not shown). Since there were no

in CDCl2CDC12 at 348 K. Signals corresponding to nongerminalled.

signals for free CONH2, eliminating asparagine and glu-tamine as possibilities, the identities of the amino acids wereconfirmed as aspartic acid, glutamic acid, leucine, andvaline. An additional low-field signal at 8 5.3 consistent withCHO of the alcohol moiety of an ester (or lactone) was alsoobserved.Attempts to obtain the sequence of the amino acids by

two-dimensional NOE (ROESY) NMR were moderatelysuccessful (Fig. 5). Germinal as well as more distant proton-proton relationships were observed in the ROSEY spectrum.These germinal proton-proton relationships, which can beconsidered as noises in peptide sequencing, were also ob-served in the TOCSY spectrum (data not shown) and,therefore, can be eliminated from the ROESY spectrum,leaving signals resulting from distant proton-proton relation-ships. These remaining proton-proton relationships yield thefollowing partial sequences that are the same as those foundin surfactin (15): fatty acid-Glu-Leu, Asp-Leu, and Leu-Leu.On the bases of the composition of the peptide moiety and

the molecular weight of the molecule, the lipid chain was

determined to be a C15 fatty acid amidated to the N-terminalamine of the peptide. It was obvious that a mixture of

Glu

Leu

Leu

Leu

-7.0

-7.2

7.4

-7.6

PPM

VOL. 60, 1994

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

APPL. ENVIRON. MICROBIOL.

CH3-CH(1)

Ante

3 1

2-CH R CH3 -CH -CH2-R1 (4) 1CH3 CH;(3) (4)

biso- Hso-

CH3 -CH2-CH2-CH2 -R(2)

Normal-

2

20 t 16 14

FIG. 6. Part of carbon NMR spectrum of the JF-2 biosurfactant.

normal, anteiso, and isobranched forms were present (CH3at 8 13.4, 10.7, and 18.7, and ca. 22, respectively) as

observed in the carbon spectrum (Fig. 6).Immunoreactivity with surfactin-specific antibodies. In or-

der to further assess the structural similarity in peptidemoiety between surfactin and the JF-2 biosurfactant, wedetermined the immunoreactivities of the JF-2 biosurfactantagainst antibodies specific to surfactin. Polyclonal antibodieswere raised against a surfactin-KLH conjugate prepared by[1-ethyl-3-(dimethylamino)propyl]-carbodiimide (EDC) cou-pling. Conjugation of the surfactin to KLH was necessarybecause surfactin (molecular weight, 1,035) alone is proba-bly too small to elicit an immune response in rabbits. Withthe surfactin-KLH conjugate, high antibody titers wereobtained. Sera were collected and used to develop an ELISAfor surfactin. The ELISA analysis showed that the poly-clonal antibodies bound to the surfactin-ovalbumin conju-gate but not to ovalbumin alone, indicating that they are

specific for the surfactin moiety of the surfactin-KLH con-jugate (data not shown).A competition ELISA was developed to demonstrate that

the polyclonal antibodies can bind free surfactin molecules.Detection by direct ELISA was not possible because freesurfactin molecules could not be applied effectively as acoating to microtiter plate wells, presumably because of theamphiphilic nature of the surfactin. In competition ELISA, asurfactin-containing sample is first preincubated with thepolyclonal antibodies and then the mixture is transferred tomicrotiter wells coated with a surfactin-ovalbumin conju-gate. The formation of antigen-antibody complexes duringpreincubation decreases the amount of free antibodies thatcan bind to surfactin-ovalbumin. The decreased amount ofbound antibodies reduces the signal resulting when the

12

washed plates are incubated with the secondary antibodies.The results of the competition ELISA are shown in Fig. 7.The absence of any color response, resulting from thecomplete blocking of the antigen binding sites on antibodiesby free surfactin molecules in the preincubation mixture inrow D, indicated that the polyclonal antibodies specificallyrecognize free surfactin. The absence of a decrease in colorresponse in control with SDS (50mg/liter in PBS) in row Bindicated that (i) the serum does not react nonspecificallywith other surface active compounds and (ii) the lack ofcolor response when the sera were preincubated with sur-factin was not due to the detergent character of the moleculecausing the desorption of the bound antigen from the wells.We found that the cyclic conformation of the peptide moietyis important for antibody binding since no competition wasobserved upon addition of a synthetic peptide with the sameamino acid sequence as surfactin (L-Glu-L-Leu-D - Leu-L -

AB

FIG. 7. The competition ELISA of surfactin and the JF-2 bio-

surfactant. (A) Control; (B) SDS; (C) the JF-2 biosurfactant; (D)surfactin. The lack of color response indicates that the binding of

antibodies to the coated surfactin-ovalbumin conjugate is inhibited.

Fifty percent serial dilutions of the samples in each row were

performed from left to right.

wm z22

36 LIN ET AL.

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

B. LICHENIFORMIS JF-2 SURFACTANT 37

100

90

. 80

~70|60

50

40-2 -1 0 1 2

Log [concentration]FIG. 8. Inhibition of binding of antibodies to coated surfactin-

ovalbumin conjugate by surfactin (-) and the JF-2 biosurfactant(0).

Val - L - Asp - D - Leu - L - Leu) (Bio-Synthesis, Inc.,Lewisville, Tex.) (data not shown).

Preincubation of the serum with a solution of the JF-2surfactant serially diluted twofold from an initial concentra-tion of 25 mg/liter gave an inhibition pattern identical to thatobtained with surfactin (Fig. 7, row C). The percent inhibi-tion by surfactin and the JF-2 biosurfactant at differentconcentrations is shown in Fig. 8. It is evident that thepercent inhibition is identical, indicating that the surfactin-specific antibodies recognize and exhibit the same affinity forthe JF-2 lipopeptide.

Interfacial activity. The effect of pH on the interfacialactivity of the JF-2 biosurfactant against decane was inves-tigated. As shown in Fig. 9, the biosurfactant preparationexhibited optimal interfacial activity at pH 6.0. The interfa-cial tensions of the solutions at pH 6.0 (0.006 and 0.023dyne/cm in the presence of 5 and 0.5% of NaCl, respectively)were 10 times lower than those at pH 7.0 with the same ionicstrength (0.060 and 0.024 dyne/cm in the presence of 5 and0.5% of NaCl, respectively). At pH 6.0, the interfacialtension in the presence of 5% NaCl (0.006 dyne/cm) was four

pHFIG. 9. The effect of pH on the interfacial activity of the JF-2

biosurfactant. Purified biosurfactant was dissolved in buffer solu-tions with 0.5% (squares) or 5% (diamonds) NaCl.

A B C DFIG. 10. The effects of microbial metabolites and NaCl concen-

tration on the interfacial activity of the biosurfactant produced by B.licheniformis JF-2. The interfacial tension against decane for solu-tions of purified biosurfactant resuspended at a concentration of 50mg/liter in fresh medium with 0.5% NaCl at pH 6.0 (A), freshmedium with 5% NaCl at pH 6.0 (B), the supernatant of acidprecipitation with 0.5% NaCl at pH 6.0 (C), and the supernatant ofacid precipitation with 5% NaCl at pH 6.0 (D) were measured.

times lower than that in the presence of 0.5% NaCl (0.023dyne/cm). A similar profile of pH effect on surfactin interfa-cial activity was also observed (data not shown). Interfacialtension measurements at each dilution of biosurfactant infreshly prepared medium with 0.5% NaCl at pH 6.0 indicatedthat the biosurfactant has a CMC of 10 mg/liter.The effects of various additives on the interfacial activity

of the biosurfactant are shown in Fig. 10. For these experi-ments, the interfacial activity of the biosurfactant was arbi-trarily defined as the reciprocal of interfacial tension againstdecane. The activity of the JF-2 biosurfactant resuspendedin the supernatant of acid-precipitated fermentation brothcontaining 5% NaCl was almost 20 times higher than that ofthe biosurfactant in freshly prepared growth medium with0.5% NaCl. This result indicates that both NaCl and acid-soluble components in the fermentation broth function ascosurfactants to augment the interfacial activity of the JF-2lipopeptide. The interfacial tension of solution containing theJF-2 biosurfactant in the supernatant of acid-precipitatedfermentation broth with 5% NaCl at pH 6.0 at a concentra-tion of 50 mg/liter was as low as 0.006 dyne/cm.

DISCUSSION

We have purified the B. licheniformis JF-2 lipopeptide byeither acid precipitation or XAD-2 adsorption chromatogra-phy as the first step followed by silica gel chromatographyand preparative-scale C18 reverse-phase HPLC. These pro-cedures gave about 25 mg of homogeneous material, asdetermined by analytical HPLC and NMR, with an overallyield of 70%.The results of chemical analysis, immunological reactiv-

ity, and NMR spectra indicate that the chemical structure ofthe peptide moiety of the JF-2 surfactant is very similar tothat of surfactin from B. subtilis. However, some differencesbetween these two biosurfactants were detected. First of all,the FAB-MS spectrum of the JF-2 biosurfactant indicatedthat the purified surfactant is a homogeneous preparation.The presence of a family of lipopeptides with the samepeptide domains but different chain lengths of lipid tails,

VOL. 60, 1994

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from

APPL. ENVIRON. MICROBIOL.

which has been frequently observed for lipopeptides pro-duced by other microorganisms (9), was not observed for thebiosurfactant produced by B. licheniformis JF-2. In addition,13C NMR analysis indicated that the lipid tail of JF-2biosurfactant is present in three different configurations,namely, n-, iso-, and anteisoforms. The anteisoform of fattyacids has not been observed in surfactin or other lipopeptidesurfactants. Nevertheless, the difference in lipid moietiesbetween surfactin and the JF-2 biosurfactant does not affecttheir interfacial activities.The effect of pH on the interfacial activity of surfactin and

the JF-2 biosurfactant cannot be easily explained by proto-nation of the carboxylic side chains of glutamic and asparticacids in the peptide moiety, since the pK. values are 4.25and 3.86, respectively. We suspected that the pH depen-dence may be a consequence of a metal chelation effectwhich alters the electrostatic properties of the molecule.However, molecular modeling by energy minimization of thecyclic peptide conformation revealed that the interatomicdistances of the carboxylic groups are inconsistent with ametal binding site either in a monomer or as a dimer (1Sa).Therefore, the pH dependence of the biosurfactant is mostlikely the result of complex electrostatic interactions at thewater/decane interface.

Results of competition ELISA with the surfactin-specificantibodies further confirmed that surfactin and the JF-2biosurfactant have exactly the same amino acid compositionand sequence. The competition ELISA can also be used forthe quantification of biosurfactant concentration in the fer-mentation culture supernatant. Detection of the biosurfac-tant by competition ELISA allows high specificity andsensitivity (at least as low as 0.01 mg/liter) as well as theability to analyze a large number of samples simultaneously.The immunoassay can also be used as a screening assay fordetecting lipopeptide-overproducing mutants of B. subtilis orB. licheniformis JF-2 (data not shown).The CMC of the JF-2 biosurfactant was determined to be

10 mg/liter, significantly lower than that of other biosurfac-tants as well as many synthetic surfactants. The CMC can beinterpreted as the solubility of surfactant in water or theminimum concentration required to reach the maximuminterfacial or surface activity. Therefore, in terms of CMC,the JF-2 biosurfactant satisfies an important criterion forcommercial detergency applications. Furthermore, weshowed that sodium chloride and metabolites in the culturesupernatant work synergistically with the JF-2 biosurfactantin reaching the maximum interfacial activity. Under optimalconditions, JF-2 biosurfactant at a concentration of 50 mg/liter reduced the interfacial tension of aqueous phase againstdecane to 0.006 dyne/cm, a 500-fold reduction from theinterfacial tension obtained in the absence of surfactant (32dyne/cm). This is the lowest value that has been reported formicrobial surfactant so far (6).

ACKNOWLEDGMENTS

Support for this work was provided by the Department of Energy,the Texas Advanced Research and Technology Program, and theCenter for Enhanced Oil and Gas Recovery Research at the Uni-versity of Texas.We are grateful to Brent Iverson for advice on producing the

surfactin conjugates and for help with computer simulations and toSandies Smith for help with amino acid analysis and Peter Zuber forproviding us with B. subtilis strains.

REFERENCES1. Arima, K., A. Kakinuma, and G. Tamura. 1968. Surfactin, a

crystalline peptidelipid surfactant produced by Bacillus subtilis:

isolation, characterization and its inhibition of fibrin clot forma-tion. Biochem. Biophys. Res. Commun. 31:488-494.

2. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G.Seidman, J. A. Smith, and K. Struhl. 1989. Current protocol inmolecular biology. John Wiley & Sons, New York.

3. Bernheimer, A. W., and L. S. Avigad. 1970. Nature and prop-erties of a cytolytic agent produced by Bacillus subtilis. J. Gen.Microbiol. 61:361-369.

4. Clark, J. B., D. Munnecke, and G. E. Jenneman. 1981. In situmicrobial enhancement of oil production. Dev. Ind. Microbiol.22:695-701.

5. Cooper, D. G., C. R. MacDonald, S. J. B. Duff, and N. Kosaric.1981. Enhanced production of surfactin from Bacillus subtilis bycontinuous product removal and metal cation additions. Appl.Environ. Microbiol. 42:408-412.

6. Georgiou, G., S. C. Lin, and M. M. Sharma. 1992. Surface-activecompounds from microorganisms. Bio/Technology 10:60-65.

7. Goursaud, J.-C. 1989. M.S. thesis. University of Texas, Austin.8. Harvey, S., I. Elashvili, J. J. Valdes, D. Kamely, and A. M.

Chakrabarty. 1990. Enhanced removal of Exxon Valdez spilledoil from Alaskan gravel by a microbial surfactant. Bio/Technol-ogy 8:228-230.

9. Horowitz, S., and W. M. Griffin. 1991. Structural analysis ofBacillus licheniformis 86 surfactant. J. Ind. Microbiol. 7:45-52.

10. Hosono, K., and H. Suzuki. 1983. Acylpeptides, the inhibitors ofcyclic adenosine 1 3',5'-monophosphate phosphodiesterase. J.Antibiot. 36:194-196.

11. Iverson, B. L., K. E. Cameron, G. K. Jahangiri, and D. S.Pasternak 1990. Selective cleavage of trityl protecting groupscatalyzed by an antibody. J. Am. Chem. Soc. 112:5320-5323.

12. Javaheri, M., G. E. Jenneman, M. J. McInerney, and R. M.Knapp. 1985. Anaerobic production of a biosurfactant by Bacil-lus licheniformis JF-2. Appl. Environ. Microbiol. 50:698-700.

13. Jenneman, G. E., M. J. McInerney, R. M. Knapp, J. B. Clark,J. M. Ferro, D. E. Revus, and D. E. Menzie. 1983. A halotoler-ant, biosurfactant-producing Bacillus species potentially usefulfor enhanced oil recovery. Dev. Ind. Microbiol. 24:485-492.

14. Jenny, K., 0. Kaeppeli, and A. Fiechter. 1991. Biosurfactantsfrom Bacillus licheniformis: structural analysis and character-ization. Appl. Microbiol. Biotechnol. 36:5-13.

15. Kakinuma, A., M. Hori, M. Isono, G. Tamura, and K. Arima.1969. Determination of amino acid sequence in surfactin, acrystalline peptidelipid surfactant produced by Bacillus subtilis.Agric. Biol. Chem. 33:971-972.

15a.Lin, S. C. Ph.D. dissertation. University of Texas, Austin.16. Lin, S. C., K. S. Carswell, M. M. Sharma, and G. Georgiou.

Continuous production of the lipopeptide biosurfactant of Ba-cillus lichenifonnis JF-2. Appl. Microbiol. Biotechnol., in press.

17. Lin, S. C., J. C. Goursaud, P. J. Kramer, G. Georgiou, andM. M. Sharma. 1990. Production of biosurfactant by Bacilluslicheniformnis strain JF-2, p. 219-226. In E. C. Donaldson (ed.),Microbial enhancement of oil recovery-recent advances.Elsevier Science Publishers, Amsterdam.

18. Lin, S. C., M. M. Sharma, and G. Georgiou. 1993. Productionand deactivation of biosurfactant by Bacillus lichenifonnis JF-2.Biotechnol. Prog. 9:138-145.

19. McInerney, M. J., M. Javaheri, and D. P. Nagle. 1990. Proper-ties of the biosurfactant produced by Bacillus lichenifornisstrain JF-2. J. Ind. Microbiol. 5:95-102.

20. Muller-Hurtig, R., R. Blaszczyk, F. Wagner, and N. Kosaric.1993. Biosurfactants for environmental control, p. 447-469. InN. Kosaric (ed.), Biosurfactants. Marcel Dekker, Inc., NewYork.

21. Oberbremer, A., R. Muller-Hurtig, and F. Wagner. 1990. Effectof the addition of microbial surfactants on hydrocarbon degra-dation in a soil population in a stirred reactor. Appl. Microbiol.Biotechnol. 32:485-489.

22. Van Dyke, M. I., H. Lee, and J. T. Trevors. 1991. Applicationsof microbial surfactants. Biotechnol. Adv. 9:241-252.

23. Wylie, D. E., D. Lu, L. D. Carlson, R. Carlson, K. F. Babacan,S. M. Schuster, and F. W. Wagner. 1992. Monoclonal antibodiesspecific for mercuric ions. Proc. Natl. Acad. Sci. USA 89:4104-4108.

38 LIN ET AL.

on February 20, 2020 by guest

http://aem.asm

.org/D

ownloaded from