J . Chem. Tech. Bivtechnol. 1990, 41, 23-29
`
`Recovery of Biosurfactants by Ultrafiltration
`
`Catherine N. Mulligan" & Bernard F. Gibbsb*
`
`"Biochemical Engineering and bProtein Engineering Sections, National Research Council
`of Canada, Biotechnology Research Institute, 6100 Royalmount Ave., Montreal,
`Quebec, Canada H4P 2R2
`
`(Received 28 December 1988; accepted 13 February 1989)
`
`ABSTRACT
`
`Ultrafiltration was used in a one-step method to purify and concentrate
`and rhamnolipids-)om
`culture supernatant
`biosurfactants-surfactin
`fluids. The ability of surfactant molecules to form micelles at concentrations
`above the critical micelle concentration allows these aggregates to be retained
`by relatively high molecular weight cut-off membranes. Lower molecular
`weight impurities such as salts, fiee amino acids, peptides and small proteins
`are easily removed. Various molecular weight cut-off membranes were
`examined for the retention of surfactin and rhamnolipids (mol. wts 1036 and
`802 respectively). Amicon X M 50 was the superior membrane for retention of
`surfactin and a I60-fold purification was rapidly achieved. The Y M 10
`membrane was the most appropriate for rhamnolipid recovery. Ultrafiltration
`can play an important role in biosurfactant purification as large volumes of
`media can be processed rapidly at extremely low cost.
`
`Key words: ultrafiltration,
`purification.
`
`surfactin,
`
`rhamnolipid, biosurfactant
`
`1 INTRODUCTION
`
`Although biosurfactants are biodegradable and very effective, commercial interest
`remains low because they are present at only low concentrations during
`fermentation.' As most of these compounds are lipid-based, classical recovery
`methods such as precipitation, crystallization and solvent extraction have been
`used.' Other methods, such as in-situ recovery, are being developed to reduce
`solvent requirement and product degradati~n.~
`* To whom correspondence should be addressed.
`23
`J. Chem. Tech. Bivrechnvl. 0268-2575/89/$03.50 0 1989 Society of Chemical Industry. Printed in Great
`Britain
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 1 of 7
`
`

`
`24
`
`C . N . Mulliyun, B. F . Gihhs
`
`HCH,CO-GLU-LEU-LEU,
`
`0 - LEU-LEU-ASP'
`
`,VAL
`
`CH3
`
`Fig. 1. Structure of B. subtilis surfactin. Amino acids are represented as: ALA-alanine, VAL-valine,
`GLU-glutamic acid, and LEU-leucine.
`
`In the case of the cyclolipopeptide surfactin, produced by Bacillus subtilis, foam
`flotation and fractionation minimize end-product inhibition and concentrate the
`surfactant. After foam collapse and cell removal, acid precipitation followed by
`solvent extraction has been used for p~rification.~ Similar recovery processes are
`required for surface active rhamnolipids R-1 and R-2 from Pseudomonas
`Adsorption and ion-exchange chromatography have also been used
`a e r ~ g i n o s a . ~ * ~
`in pilot-plant studies.' These are key examples of solvent- and labour-intensive
`processes.
`In this study, ultrafiltration was evaluated as a method to concentrate surfactin
`(Fig. 1) and rhamnolipids from the collapsed foam. Surfactant molecules form
`micelles at concentrations higher than the critical micelle concentration (CMC),
`and the remaining molecules remain unassociated.'-" Micelles would be retained
`by high molecular weight cut-off membranes.
`Biosurfactants have very low CMCs, ideal for ultrafiltration. The CMC of the
`cyclolipopeptide has been reported4 as 0.025 g dmT3 with a molecular weight of
`1036." The rhamnolipids have CMCsl3 of 0.050-0.200 g d m - j with molecular
`weights for R-1 and R-2 of 744 and 802, r e s p e ~ t i v e l y . ~ ~ ~
`Various molecular weight cut-off membranes were evaluated for their ability to
`concentrate and purify the biosurfactants.
`
`2 MATERIALS AND METHODS
`
`2.1 Microorganisms
`Bacillus subtilis ATCC 21332, was maintained at 4°C on 4 % glucose, mineral salt
`medium4 agar plates. Pseudomonas aeruginosa ATCC 9027 was maintained on
`Pseudomonas agar P (Difco).
`2.2 Cultivation conditions
`After 3 days growth in 100cm3 of 4% glucose and mineral salt medium
`supplemented with 3 . 2 ~ mol dm-3 FeSO,, 50cm3 of B. subtilis was
`transferred into 500 cm3 (2 dm3 flask). After 6 h of growth, inoculum (0.5 dm3
`10.0 dm-3) was added to a 20 dm3 Bioengineering fermenter. The following
`cultivation conditions were used: aeration at 20dm3 min-I, pH control at 6.7,
`100 rpm agitation and 37°C. Foam was collected and collapsed in a flask on the air
`exhaust line.4 Pseudomonas aeruginosa was grown in a similar manner in proteose
`peptone m e d i ~ m . ' ~ Cells were removed by centrifugation in a Beckman centrifuge
`at 12 OOOg for 10 min.
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 2 of 7
`
`

`
`2.3 Ultrafiltration
`Cell-free foam fractions were concentrated by an Amicon magnetically stirred
`ultrafiltration cell, containing a YM 10, YM 30, XM 50, XM 100 or XM 300
`membrane (mol. wt cutoffs of 10 000, 30 000, 50 000, 100 000 and 300 000 daltons
`respectively). A pressure of 172 kPa was used.
`
`2.4 Analytical methods
`Surface tension, CMC, and amino acid concentration were determined on the
`permeate and the retentate throughout ultrafiltration. Surface tension was
`determined by the de Nouy method with a Fisher Tensiomat Model 21. The CMC
`was determined by measuring the surface tension at various dilutions.' ' The
`logarithm of the dilution was plotted as a function of the surface tension. The CMC
`is the point at which the surface tension abruptly increases. The reciprocal of CMC
`is an indication of relative concentration.
`The amount of surfactin was determined by amino acid analyses. A 10pm3
`aliquot was dried and acid hydrolysed for 2.5 h at 150°C in a PICO-TAG amino
`acid analysis system. The residue was redissolved in 200 pm3 of sodium buffer and
`injected on a Beckman System 6300 high performance analyser equipped with a
`Beckman Model 7000 data station. All buffers and ninhydrin reagents were
`purchased from Beckman. The concentration of surfactin was calculated by
`multiplying the lipopeptide concentration (mol dm-3) by the molecular weight
`( I 036).
`2.5 Chemical isolation of surfactin
`Surfactin was isolated by adding concentrated hydrochloric acid to the collapsed
`foam after cell r e m ~ v a l . ~ Dichloromethane (1 :I, v/v) was added to the suspension in
`a separatory funnel and shaken vigorously. The aqueous (bottom) layer was
`removed and extracted twice more as described above. The organic layer was
`pooled and evaporated. The residue was redissolved in water (pH 8.0) and filtered
`through Whatman No. 1 paper to remove undissolved impurities. Concentrated
`HCl was again added to the filtrate and extracted with dichloromethane ( I :1, v/v)
`three times and evaporated as described.
`2.6 Viscosity measurement
`The viscosity of the surfactin solution was measured by the Couette principle using
`a Contraves Low Shear 30 rotational rheometer. Temperature was controlled at
`26°C by a Contraves Rheotherm 115 water bath.
`
`2.7 Glucose and phosphate measurements
`Glucose and inorganic phosphates were analysed using a Waters high performance
`liquid chromatograph (HPLC), equipped with a Digital Model 350 computer. For
`the glucose analysis, a Shodex DC613 column was used. With a mobile phase of
`.acetonitrile-water (70:30) at a flow rate of 0.8 cm3 min-' (50"C), glucose was
`detected by a Waters 401 differential refractometer.
`Phosphates were detected by a Waters 430 conductivity detector, mobile phase of
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 3 of 7
`
`

`
`26
`
`C. N. Mulligun, B. F . Gihbs
`
`ghconate (16 g dm-'), boric acid (18 g dm-'), and sodium tetraborate. 1OH,O
`(25 g dm-'), at a flow rate of 075 cm3 min- ' (45°C). The analyses were performed
`using a Waters ICPAK A column.
`2.8 Massspectrometry
`Mass spectra were obtained in the positive ion mode on a VG Analytical ZAB-HS
`double focusing mass spectrometer. The accelerating voltage was 10 kV and the fast
`xenon atom beam was operated with an emission current of 1 mA at 8 kV. Mass
`spectra were recorded with an integrated data acquisition system and calibration
`was performed with CsI. Spectra for samples are an average of 10 scans.
`
`3 RESULTS AND DISCUSSION
`
`Five different ultrafiltration membranes were evaluated for their ability to retain
`surfactin and remove impurities. A 10cm3 sample of collapsed foam was
`concentrated to c. 1 cm3 in each case. The results are shown in Table 1. Surfactin
`retention by the three lower molecular weight cut-off membranes was superior to
`the remaining two. This implies that between 50 and 100 molecules aggregate to
`form micelles. The permeates (YM 10, YM 30 and XM 50) contained only the
`surfactin molecules which were unassociated. This was confirmed by the relatively
`high surface tension measurements. Because of the low CMC of the surfactant, only
`small fractions of the molecules pass through these membranes. Although the CMC
`has been reported4 as 0025 g dm-3, a concentration of 0.01 1 g dm-3 was
`determined by amino acid analysis and surfactant dilutions.
`Threonine, serine, glycine and alanine were found in the foam fraction, in
`addition to the surfactin amino acids. Retention of these impurities decreased as the
`molecular weight cut-off increased (Table 1 ). Although ultrafiltration improved the
`surfactin amino acid composition, no significant difference was seen between the
`membranes.
`
`TABLE 1
`Purification of E. subtilis Surfactin by Ultrafiltration
`
`Membrane Retention of
`surfitin
`(%I
`
`Surfactin
`amino ucids
`in retentate
`(%)
`
`Purification
`,fuctor
`
`Viscosity of
`retentutes
`( C P )
`
`Surjuce
`tension of
`permeates
`(mN m-')
`
`None"
`YM 10
`YM 30
`XM 50
`XM 100
`XM 300
`
`0.0
`98.2
`%.8
`98.2
`73.8
`28.0
`
`92.9
`96.6
`96.7
`96.9
`97.1
`97.6
`
`I .o
`9.9
`9.9
`9.8
`7.8
`2.9
`
`I .O
`1.2
`1.3
`1.2
`1.2
`1 .0
`
`27.8
`34.6
`31.7
`31.9
`3@5
`30.6
`
`a Data represent the characteristics of the surfactin solution before ultrafiltration.
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 4 of 7
`
`

`
`Recovery of hiosur/acrunrs by ultrujltration
`
`21
`
`The retention of other impurities by the membranes was also verified. Glucose
`(10 g dm-3) and inorganic phosphate (4.2 g dm-') were two major components in
`the growth medium. None of the glucose and 10% of the phosphates were retained
`by each of the membranes. Ultrafiltration efficiently removed free amino acids,
`small peptides, proteins and medium components from the product.
`In a larger-scale experiment, 7 dm3 of collapsed foam was passed through a Y M
`30 membrane. A purification factor of 160 was achieved with 90% retention of the
`surfactant. A surfactin concentration of 51-8 g dm-3 was obtained. The purity
`(52.6 %)of the dried surfactin was superior to that of the chemically purified product
`(3 1.6 %). Mass spectrometry (Fig. 2) confirmed the structure of the surfactin. The
`spectrum agrees with the original identification of the compound.'* The protonated
`molecular ion ( M W + H + ) is seen at M/Z= 1037.
`Concentrating surfactin up to 10-fold (Table 1) did not significantly increase
`viscosity. However, as the surfactin concentration increased by 160-fold, the
`viscosity became significant (4.1 CP at 51.8 g d n ~ - ~ ) , retarding the filtration rate. In
`
`view of the above, surfactin concentrations should be limited to ca. 20g dnC3.
`Retention of rhamnolipids from P. aeruginosa by ultrafiltration membranes was
`examined. The YM 10 (Table 2) was the most effective membrane as only a small
`fraction of the biosurfactant passed through this membrane. These rhamnolipid
`micelles are smaller than the surfactin aggregates.
`In summary, ultrafiltration is a simple technique for surfactin purification.
`Membranes with relatively high molecular weight cut-offs (i.e. 50 000) can be used
`as aggregates of 5&100 molecules are formed at a concentration above the CMC.
`Impurities can be easily removed with minimum surfactin loss. Recovery costs are
`dramatically reduced, as large volumes of solvents are not required. In addition, this
`method requires only a fraction (ca. 2%) of the time required for the quickest
`previously published m e t h ~ d . ~ This technique is not restricted to lipopeptide and
`
`Mo
`
`>
`k s o
`In z
`W
`I-
`z 6 Q
`
`i
`y 4 l l
`
`K 520
`
`0
`1000
`
`1020
`
`1040
`1060
`M / Z
`Fig. 2. Mass spectrum of purified B. suhtilis surfactin. The protonated molecular ion is seen at 1037. The
`spectrum agrees with the original authors.12
`
`1080
`
`1100
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 5 of 7
`
`

`
`28
`
`C . N. Mulliyrtn, B. F . Gihhs
`
`TABLE 2
`Purification of P. aeruginosa Rhamnolipid by Ultraliltration
`
`Membrane
`
`Retention of
`surjactant
`(%I
`
`Purification
`factor
`
`None"
`YM 10
`YM 30
`XM 50
`XM 100
`XM 300
`
`0.0
`92.0
`80.0
`58.9
`40.0
`22.2
`
`1 .o
`9.2
`8.0
`4.8
`2.6
`2.0
`
`Viscosity oj'
`retentates
`(CP)
`
`Surfiice
`tension 41'
`permeates
`(mN m-')
`
`0.93
`I .46
`1.53
`1.27
`1.25
`1.14
`
`28.9
`31.3
`31.2
`29.8
`29.9
`29.8
`
`Data represent characteristics of surfactant solution before ultrafiltration.
`
`rhamnolipid biosurfactants but can also be used for molecules that tend to
`aggregate above certain concentrations.
`
`ACKNOWLEDGEMENT
`
`The authors gratefully acknowledge Dr Orval A. Mamer of the McGill University
`Biomedical Mass Spectrometry Unit for his expertise in mass spectrometry.
`
`REFERENCES
`
`1. Cooper, D. G., Biosurfactants. Microbiol. Sci., 3 (1986) 145-50.
`2. Syldatk, C. & Wagner, F., Production of biosurfactants. In Eiosurjuctunts und
`Biotechnology, ed. N. Kosaric, W. L. Cairns & N. C. C. Gray. Marcel Dekker, New
`York, 1987, pp. 89-120.
`3. Rofller, S. R., Blanch, H. W. & Wilke, C. R., I n situ recovery of fermentation products.
`Trends Biotechnol., 2 (1984) 129-36.
`4. Cooper, D. G., MacDonald, C. R., Duff, S. J. B. & Kosaric, N., Enhanced surfactin
`production from Bacillus subtifis by continuous product removal and metal cation
`addition. Appl. Enuiron. Microbiol., 42 (1981) 408-12.
`5. Hirayama, T. & Kato, I., Novel methyl rhamnolipids from Pseudomonus ueruginosu.
`FEES Lett., 139 (1982) 81-5.
`6. Itoh, S., Honda, H., Tomita, F. & Suzuki, T., Rhamnolipid produced by Pseudomonus
`aeruginosa grown on n-paraffin. J . Antibiot., 24 (1971) 855-9.
`7. Reiling, H. E., Thanei-Wyss, U., Guerra-Santos, L. H., Hirt, R., Kappeli, 0. & Fiechter,
`A., Pilot plant production of rhamnolipid biosurfactant by Pseudomoniis aeruyinosir.
`Appl. Enuiron. Microbiol., 51 (1986) 985-9.
`8. Rosen, M. J., Surfactants and Interfacial Phenomena. Wiley, New York, 1989, pp. 108-
`69.
`9. Shinoda, K., Colloidal Surjactunts, ed. K. Shinoda, T. Nakagawa, B. Tamamushi and T.
`Isemura. Academic Press, New York, 1963, pp. 1-96.
`10. Mysels, K. J., Charge effects in light scattering by association colloids electrolytes. J .
`Colloid Sci., 10 (1955) 507-22.
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 6 of 7
`
`

`
`Recovery q/' biosw/ucrurirs by irltrufiltruriorr
`
`29
`
`11. Mukerjee, P., The nature of the association equilibria and hydrophobic bonding in
`aqueous solution of association colloids. Adv. Colloid lnterfuce Sci., 1 (1967) 241-75.
`12. Kakinuma, A., Oachida, A,, Shina, T., Sugino, H., Isono, M., Tamura, G. & Arima, K.,
`Confirmation ofthe structure by mass spectrometry. Ayric. Biol. Chem., 33 (1969) 1669-
`71.
`13. Wagner, F., Kim, J.-S., Lang, S., Li, Z.-Y., Marwede, G., Matulovic, U., Ristau, F. &
`Syldatk, C., Production of surface active anionic glycolipids by resting and immobilized
`microbial cells. In Proc. Third European Congress, Biotech., Vol. 1, ed. DECEMA.
`Verlag Chemie, Weinheim, 1984, pp. 1-3-1-8.
`14. Cheng, K. J., Ingram, J. M. & Costerton, J. W., Release of alkaline phosphatase from
`cells of Pseudomonas aeruginosa by manipulation of cation concentration and pH.
`J . Bacteriol., 104 (1970) 748-53.
`15. Cooper, D. G., Zajic, J. E. & Gerson, D. F., Surface active compounds from
`microorganisms. Adv. Appl. Microbiol., 26 (1979) 229-53.
`
`PETITIONERS
`
`EXHIBIT NO. 1013 Page 7 of 7