Recovery of Thuringiensin with Cetylpyridinium Chloride UsingMicellar-Enhanced Ultrafiltration Process

Yew-Min Tzeng,*,† Hung-Yieng Tsun,† and Yaw-Nan Chang‡

Institute of Biotechnology, National Dong Hwa University, Shoufeng, Hualien, Taiwan 974, ROC, andDepartment of Food Engineering, Da-Yeh University, Dahtsuen, Changhwa, Taiwan 515, ROC

A cationic cetylpyridinium chloride (CPC) surfactant-based micellar-enhanced ultra-filtration (MEUF) process was used to recover thuringiensin from the supernatant ofBacillus thuringeinsis fermentation broth. Several different strategies of manipulatingprocess variables, such as ionic strength, pH, CPC concentration, micelle formationtemperature, and membrane pore size, were investigated. It was found that CPCconcentration and membrane pore size were two major factors to the increase ofthuringiensin recovery, while the ionic strength and pH adjustments were notnecessary and micelle formation temperature was not important within the temper-ature range studied. Finally, a novel two-step MEUF processing scheme was developedwhich includes microfiltration and two-step ultrafiltration (UF) processes. Thethuringiensin recovery up to 94.6% was attained when an initial CPC concentrationof 4% (w/v) and UF membrane of molecular weight cutoff (MWCO) of 3000 was usedin conjunction with the previous UF process using a MWCO 30 000 membrane. Thebioassay results showed that the spray-dried thuringiensin with CPC was more toxicto fly larvae than without CPC. This research demonstrated that CPC not onlyfacilitated thuringiensin recovery but also improved the insecticidal effect.

Introduction

Thuringiensin is one of eight insecticidal toxins pro-duced by a number of strains of Bacillus thuringiensis ,which is a low molecular weight of 701, thermostabletoxin known as â-exotoxin (Sebesta et al., 1981). Thur-ingiensin (C22H32N5O19P1‚3H2O shown in Figure 1) isknown as a nucleotidic ATP analogue (Benz, 1966;Campbell et al., 1987) that inhibits the production ofDNA-dependent RNA polymerase and consequently in-terferes with the production of ribosomal RNA by com-peting with ATP for binding sites (Lecadet and de Barjac,1981). It is observed that thuringiensin is toxic to a broadrange of insects, including Diptera, Coleoptera, andLepidoptera (Krieg and Langenbruch, 1981). Meanwhile,it is considered an effective means of controlling thelarval development and maturation, adult longevity, andfecundity of flies (Ignoffo and Gregory, 1973) and isreferred to as the “fly factor”. Thuringiensin also resultsin deformities, particularly during critical developmentalstages such as molting, pupation, and metamorphosis.

Thuringiensin was first purified from aqueous super-natant fermentation liquor by adsorption using ionicexchange resin, as reported by de Barjac et al. (1966).However, the recovery was low, ca. 15% only. Otherprocesses such as evaporation at 50 °C for concentratingthuringiensin (Benz, 1966), salt precipitation of thuring-iensin (by addition of calcium, magnesium, barium salts,etc.), and some other modified methods (Kim and Hung,1970) were developed in later years, but the recoverieswere not satisfactory. Another drawback to the recovery

of the thuringiensin from the supernatant of B. thuring-iensis fermentation broth by using traditional methodsis lack of efficiency. Recently, Tzeng and Hsu (1994)developed a separation method for thuringiensin byextraction in a poly(ethylene glycol) (PEG)/dipotassiumphosphate (K2HPO4) two-phase aqueous system. Therecovery was improved up to 84% by using PEG 8000and 10% K2HPO4. Although the PEG/salt system is veryeffective in the recovery of thuringiensin from thesupernatant without losing activity, the process is time-consuming to allow separation of the two phases. Tra-ditional ultrafiltration (UF) is ineffective in removingdissolved low molecule weight organics from water.Micellar-enhanced ultrafiltration (MEUF) is a recentlyproposed technique to separate dissolved organic com-pounds from aqueous streams (Dunn et al., 1985; Bhatet al., 1987; Gibbs et al., 1987; Christian et al., 1988). Inthis process, surfactant is added to an aqueous streamcontaining organic solute for forming micelles in orderto separate target compound for subsequent furtherpurification. This surfactant-based MEUF separation

* To whom correspondence should be addressed.† National Dong Hwa University.‡ Da-Yeh University.

Figure 1. Chemical structure of thuringiensin.

580 Biotechnol. Prog. 1999, 15, 580−586

10.1021/bp990053o CCC: $18.00 © 1999 American Chemical Society and American Institute of Chemical EngineersPublished on Web 05/06/1999

process is emerging as a major class of unit operationfor industrial separations.

One important point of the surfactant-based MEUFprocess is the promotion of surfactant aggregate forforming micelles to effectively adsorb the target com-pound. Several operation factors can be manipulated tocause an increase in micelle formation and aggregationfor further separation and purification. Some of thesefactors are the pore size of UF membrane, the type andconcentration of surfactant, the set of experimentalconditions (such as pH, ionic strength, temperature, andtrans-membrane pressure, etc.), and so on (Dunn et al.,1985; Pramauro et al., 1992, 1993).

In this research, the recovery of thuringiensin byseparation in the MEUF process was investigated byusing a cationic surfactant cetylpyridinium chloride(CPC). The effects of ionic strength, pH manipulation,micelle formation temperature, membrane pore size, andCPC surfactant concentration were investigated. Finally,the fly bioassay of the spray-dried powder of thuring-iensin with CPC addition, the filtrated supernatant ofthuringiensin obtained from B. thuringiensis fermenta-tion broth, and the thuringiensin standard were com-pared.

Materials and MethodsMicroorganism and Chemicals. B. thuringiensis

subsp. darmstadiensis (HD-199) was obtained from Dr.de Barjac, Institute Pasteur, Paris, France. Maintenanceand storage of the strain HD-199 were conducted accord-ing to Tzeng and Young (1995). All chemicals used wereof reagent-grade quality.

Thuringiensin Fermentation. The soy molassesbroth (SMB) medium was used in the fermentationsaccording to Tzeng and Young (1996). A loopful of thestrain HD-199 from a nutrient agar plate was inoculatedinto a 1-L Erlenmeyer flask with 250 mL of the precul-ture medium containing 5 g of yeast extract (Difco) and8 g of nutrient broth (Difco) per liter and was cultivatedat 30 °C on a rotary shaker (Hotech, Taiwan) at 200 rpmfor 12 h. The net-draft-tube modified airlift fermenter wasused in this study. The equipment and experimentalsetup of the fermenter were described previously (Jonget al., 1995; Tzeng and Young, 1996; Tzeng and Chang,1996). After the SMB medium in the fermenter wassterilized, an inoculum (10% of the working volume) wastransferred from the flask of the seed culture to thefermenter containing 13 L of the initial liquid volume.The fermentation was conducted for 44 h at 30 °C. ThepH was controlled at 7.0 using 3 M H2SO4 and 6 N NaOH.The air flow rate was controlled at 1.5 vvm. At the endof the fermentation, the culture broth was sterilized at121 °C and 101 kPa for 20 min. After cooling, the brothwas centrifuged at 8820g (J2-21M/E, Beckman) for 15min to collect the supernatant. The supernatant was thenfiltered by a cellulose acetate membrane filter (0.45 µm,Gelman) for later thuringiensin analysis and recovery byusing MEUF. The pH of the filtered supernatant was6.8-7.0 without any adjustment.

Thuringiensin Analysis. The analysis of thuring-iensin was done according to Campbell et al. (1987),Levinson et al. (1990), Wu et al. (1993), and Liu andTzeng (1998) with some modifications. The thuringiensinconcentration was determined by HPLC assays. Theinstrument of the HPLC system (HPLC 10AS, Shimadzu)included a LC 9A pump, a UV/vis SPD-6A detector setat 260 nm, and a 6A integrator. The assays wereperformed isocratically on a Waters µ-Bondapak C18column (30 cm, 3.9 mm i.d.) maintained at room tem-

perature. The mobile phase was 50 mM KH2PO4 (the pHwas adjusted to 2.8 with H3PO4) and was pumped intothe column at a flow rate of 1.0 mL/min. A standard curvewas established using a thuringiensin standard (providedby Dr. de Barjac); the thuringiensin concentration wascalculated on the basis of the HPLC results and thecalibration curve.

Thuringiensin Recovery. Filtered supernatant, whichcontained thuringiensin, was mixed with CPC, and thenthe thuringiensin recovery was conducted by MEUF. TheMEUF process is shown in Figure 2. Cellulose acetatemembranes (Amicon, YM-30, YM-10, and YM-3), 62 mmi.d., were used in the UF study. Effective areas of all ofthe membranes were 28.7 cm2. The molecular weightcutoffs (MWCOs) of the membranes YM-30, YM-10, andYM-3 were 30 000, 10 000, and 3000, respectively. TheUF experiments were carried out in 200-mL stirred cells(Model 8200, Amicon). The pressure drop across themembrane was maintained at 207 kPa during each run.The cell was initially filled with 150 mL of the thuring-iensin/CPC mixture. The mixture was agitated at 500rpm for 5 min, and then, the mixtures were allowed toequilibrate at room temperature (20-30 °C) for 30 min.The UF was operated until only 20 vol % was retainedin the cell (30 mL of concentrated solution as theretentate). The permeate was collected, and liquid fluxeswere measured. The permeate was then analyzed forthuringiensin concentration by HPLC as described above.The thuringiensin recovery (%) based on the amount ofthuringiensin in the concentrated solution in the reten-tate and that before microfiltration (MF) and UF wascalculated. Several operation variables were investigatedin this study:

(1) Effect of Ionic Strength. After 0.45-µm mem-brane filtration, the supernatant samples were addedwith the inert salt (NaCl) to reach different final con-centrations (0, 1, 2, 3, 4, 5, and 6% w/v). The sampleswere then mixed with CPC, and the final concentrationof CPC was 3% (w/v). The YM-30 (MWCO 30 000)membranes were used in the UF processes. Filtrationwas conducted at a pH of 6.8-7.0 and room temperature.

(2) Effect of pH. The filtered supernatant was thenmixed with CPC, and the final concentration of CPC was4% (w/v). The pH of the mixtures was adjusted todifferent values (6.8, 7.0, 7.5, 8.0, 8.5, and 9.0) by usingNaOH and HCl solutions. The mixtures were filtered byusing YM-30 membranes in the UF processes at roomtemperature.

Figure 2. Schematic diagram of MEUF process for thuring-iensin/CPC micelle formation and separation.

Biotechnol. Prog., 1999, Vol. 15, No. 3 581

(3) Effect of Surfactant Concentration and Mem-brane Pore Size. The CPC concentration in the filteredthuringiensin supernatant without pH adjustment wascontrolled at various concentrations (0.5%-4.0%), and thepH of this mixture was about 7.0 without ionic strengthadjustment. To study the effect of the pore size of UFmembrane, both the YM-30 and YM-10 (MWCO 10 000)membranes were used in the UF process and compared.The UF processes were carried out at room temperature.

(4) Effect of Micelle Formation Temperature. Thu-ringiensin solutions of different CPC concentrations (4%and 8%) without pH and ionic strength adjustment wereagitated at 500 rpm for 5 min. The solutions in the stirredcells were kept at different temperatures (25, 45, 50, 55and 60 °C) in an oven for 30 min in order to equilibratethe micelle formation. After equilibration, the samplesin the cells were cooled to room temperature, followedby UF. Both the YM-30 and YM-10 membranes wereused in the UF process. The effects of temperature, CPCconcentration, and membrane pore size on thuringiensinrecovery were studied.

(5) Effect of Two-Step MEUF Process. To improvethe one-step MEUF process, which was energy consum-ing, as described above in (4) with micelle formationunder controlled temperature, the two-step MEUF pro-cess was designed to operate at room temperature.Thuringiensin solutions of different CPC concentrations(4% and 8%) without pH adjustment and the YM-30membranes were used in the first step of this process.After the first-step MEUF process, the permeated solu-tions were then mixed with CPC to reach different finalconcentrations (1, 2, 3, and 4%). For further study theeffect of MWCO of UF membranes, both YM-10 and YM-3(MWCO 3000) membranes were used in the second-stepUF process. The second permeate was then analyzed forthuringiensin concentration by HPLC as described above.The final retentate was obtained from the second reten-tate mixed with the first retentate in this two-stepprocess. The thuringiensin recovery (%) for the finalretentate was calculated as described previously.

Scale-Up of Two-Step MEUF Process. To scale-upthe processes of the two-step MEUF, a 2000-mL hollowfiber concentrator system (Model CH2PRS1 for H1 hollowfiber cartridge use, Amicon) was used in the UF experi-ments operated at room temperature. The cartridge wasoperated at 69 kPa during each run. Polysulfone hollowfibers (Amicon, H1P10-43, H1P3-20), 20.3 cm in length,were used in the UF study. The surface areas of theH1P10-43 and H1P3-20 fibers were 0.03 and 0.06 m2,respectively. The 2000-mL reservoir was initially filledwith 1000 mL of the thuringiensin/CPC mixture, andthen the mixtures were allowed to equilibrate at roomtemperature for 30 min. The recirculation rate wascontrolled at 0.8 L/min. The UF was operated until only20 vol % (200 mL of concentrated solution) was retainedin the reservoir. Thuringiensin solutions of 4% of CPCconcentration without pH adjustment and the H1P10-43 (MWCO 10 000) fiber were used in the first step ofthis process. After the first-step MEUF process, the 800-mL first permeated solution was then mixed with CPCto reach the final concentration of 4%. The H1P3-20(MWCO 3000) fiber was used in the second step, and therecirculation rate was also maintained at 0.8 L/min. Thesecond permeate was then analyzed for thuringiensinconcentration by HPLC. The UF was operated until only20 vol % was retained in the retentate side (160 mL ofconcentrated solution in the second retentate). The finalretentate was obtained, and the thuringiensin recovery(%) was calculated. The final retentate of concentrated

thuringiensin/CPC mixture was dried with a spray dryer(Model SD-2, EYELA, Japan). The spray-dried powderof thuringiensin with CPC was for later use in thebioassay test.

Fly Control Bioassay. An agar-based diet modifiedfrom that reported by Ignoffo and Gard (1970) and Hsuet al. (1997) was used to assay thuringiensin activityagainst 3-day-old larve of Musca domestica (supplied byDr. Suey-Sheng Kao, Taiwan Agricultural Chemicals &Toxic Substances Research Institute, Taiwan, ROC). Themodified diet consisted of 7 g of bacto-agar (Difco), 36 gof powdered skimmed milk (commercial), and 186 g ofyeast powder (Taiwan Sugar Co.) per liter. The spray-dried powder of thuringiensin with CPC was mixedthoroughly with a little hot agar solution, and thethuringiensin concentration in the diet was exactlycontrolled at 30 °C for later use in the assay. Ten larvaewere transferred into each Petri dish containing 15 g ofthe solidified diet of various thuringiensin levels and wereincubated at 30 °C. The numbers of larvae, pupae, normaladults, and deformed adults were counted after 1, 2, 3,4, 5, and 10 day(s) for activity measurements. Thenumbers at each test level were measured in quadrupli-cate. Nonmolted pupae were counted as dead. Activitymeasurements were based upon mortality after 5 daysand adult emergence after 10 days. Deformed flies atadult stage were counted as nonemerged.

Results and DiscussionsThuringiensin Recovery. The selection of a surfac-

tant is important in designing separation processes basedon MEUF (Kandori and Schechter, 1990). As reportedby Tsun and Tzeng (1994), the recovery of thuringiensincould be improved by using cationic surfactants such ascetylpyridinium chloride and cetyltrimethylammoniumchloride. Owing to their ionic character, these cationicsurfactant aggregates could exert an electrostatic attrac-tion interaction toward the thuringiensin anions. Thisled to a presumption that thuringiensin could be nega-tively charged in the pH range 6.8-7.0 and was capableof binding with cationic surfactants in micelle aggregates.To improve the performance of MEUF, it is necessary toincrease the amount of thuringiensin to bind with micelleaggregates by introducing additional interactions. In anattempt to achieve this, the effects of ionic strength, pH,CPC concentration, membrane pore size, and micelleformation temperature were investigated.

The effect of ionic strength on thuringiensin recoverywas first studied. Figure 3 shows the recoveries in MEUFexperiments could be significantly decreased by addingof NaCl, due to neutralization of the micelle charge (Issidet al., 1992). In this sense, the ionic micelle could beregard as a neutral aggregate. Adding NaCl salts shouldreduce thuringiensin solubilization and the electrostaticinteraction between thuringiensin and CPC micelles. Forthe thuringiensin/CPC system, where the electrostaticattraction interaction significantly decreased the solute’sbinding to the aggregates, the recovery was lower athigher ionic strength. Second, the effect of the adjustedpH (up to 9.0) had been investigated. Figure 4 shows nosignificant change in the value of thuringiensin recoveryobserved in MEUF experiments conducted at high pHwith cationic micelles, and this could be attributed to thegood electrostatic attraction interaction between thur-ingiensin anions and the positively charged aggregatesat pH 6.8-9.0. It appears that the pH effect is notparticularly important for the thuringiensin recovery,where the electrostatic contribution is more relevant forthuringiensin/CPC micelles having larger interactionbinding in close neutral or more alkali solutions.

582 Biotechnol. Prog., 1999, Vol. 15, No. 3

Third, the effects of surfactant concentration andmembrane pore size (or MWCO) were examined. As canbe seen from Figure 5, the thuringiensin recovery in-creased with the initial CPC concentration in the cell upto a saturation limit. Increasing the concentration of theCPC surfactant should increase the ion-binding abilityof thuringiensin because of the electrostatic attractioninteraction or charge effect. However, the tendency ofthuringiensin to bind the micelle aggregates would bereduced as CPC concentration increased up to a certainlevel due to the ion-binding plateau or electrostaticinteraction maximum. The effect of the MWCO of themembrane on recovery is also shown in Figure 5. Therecoveries for a given high CPC concentration (g3% w/v)were a little different for the two membranes. A slightdifference in the characteristics of the MWCO of UFmembranes was seen in the recovery (blocking) of micelle-forming substances at high concentration. The resultsshow slightly higher permeate thuringiensin concentra-tions when using the 30 000 MWCO membrane thanusing the 10 000 MWCO membrane, consistent with adecrease in thuringiensin/CPC micelle rejection. Theresult is in close agreement with that reported by Dunnand co-workers (1987). It would be due to the expectedlower intrinsic rejection of the larger pore size membrane.Therefore, at high CPC concentration, the recovery with

the larger MWCO was slightly lower than with thesmaller MWCO. However, the difference in recovery wasless significant at low concentrations because the twomembranes were not capable of blocking the smallermicelle aggregates. Actually, the thuringiensin recoverynot only depends on surfactant concentration and mem-brane MWCO but also on micelle formation. Further-more, the micelle formation and micelle sizes were foundto be strongly influenced by temperature (Hayashi andIkeda, 1980; Missel et al., 1980; Os et al., 1991). Theoptimal temperature range for micelle formation shouldbe determined. Under controlled temperature the mi-celles were allowed to form at 25, 45, 50, 55, and 60 °C,and then the whole system was cooled to room temper-ature for micelle aggregation, followed by UF separation.The results show that the optimal temperature formicelle formation may be in the range 45-55 °C (Table1a,b). The electrostatic interaction between thuringiensinand CPC would become weak, and the micelles shouldbreak up into monomers or shift to small ones at hightemperature (Missel et al., 1980; Os et al., 1991). How-ever, the electrostatic interaction should stablize themicelle aggregations and induce the growth of largeaggregates at room temperature, which facilitated theUF process. Unfortunately, the results show that control-ling the micelle formation temperature at 25-60 °C inMEUF processes using the two different MWCO mem-branes did not significantly improve the thuringiensinrecoveries (for both 4% and 8% of initial CPC concentra-tion, Table 1a,b). The results also show that the recoveryincreased with increasing the initial CPC concentrationand was higher when a lower MWCO membrane wasused. Therefore, it can be concluded that for differentoperation conditions (CPC concentration and membranepore size) the controlled temperature did not significantlyaffect the final results of thuringiensin recovery.

Taking the above observations into account (the con-trolled temperature of micelle formation was not impor-tant), the two-step MEUF process without intense energyconsumption was designed to operate at room tempera-ture. Radke and co-workers (1995) have suggested that,if the permeate solute concentration cannot be reducedsufficiently by a single ultrafilter, additional surfactantmay be added to the permeate, which is filtered again.The solute concentration in the permeate from the secondultrafilter is thereby reduced further (Markels et al.,1995a). On the basis of the above suggestion, the 3000and 10 000 MWCO membranes were selected to use asthe second filter membrane in the two-step MEUF

Figure 3. Effect of NaCl salt concentration (0-6% w/v) in thefiltered supernatant on thuringiensin recovery (%) in MEUFprocess; the initial CPC concentration was 3% (w/v).

Figure 4. Effect of pH of the filtered supernatant (pH range6.8-9.0) on thuringiensin recovery (%) in MEUF process; theinitial CPC concentration was 4% (w/v).

Figure 5. Effect of MWCO of UF membrane on thuringiensinrecovery (%) in MEUF process for different initial CPC concen-trations (0.5-4.0% w/v).

Biotechnol. Prog., 1999, Vol. 15, No. 3 583

process and compared the membrane performance in thesecond step of the process. The thuringiensin recoverieswere 70.2% and 78.1% for CPC initial concentrations of4% and 8%, respectively, using the MWCO 30 000membranes in the first step of the two-step process. Table2a,b shows that the recoveries increased as the concen-tration of CPC was increased in the second step of theprocess. Thuringiensin recovery further increased whena lower MWCO membrane, such as YM-3 (MWCO 3000)membrane, was used (Table 2a,b). When initial CPCconcentration was 4% in the second step, the recoveriesof the final retentate were 92.9% and 94.6% using theMWCO 10 000 and 3000 membranes, respectively (partsa and b of Table 2, respectively). The above observationsled to a conclusion that CPC concentration and lowmembrane MWCO were major factors for high thuring-iensin recovery.

Two-Step MEUF Process. A processing scheme(Figure 6) for the recovery of thuringiensin was developedwhich included MF using a 0.45-µm acetate filter andtwo steps of MEUF processes using different MWCOhollow fibers with the additions of CPC surfactant.Markels et al. (1994, 1995b) recently reported that thepermeation flux through the membrane with the largerMWCO was higher than that with the smaller MWCOand that the flux decreased with increasing the surfac-tant concentration in the retentate because the resistancethrough the precipitate layer was increased due to bulkyaggregates. The scale-up process of two-step MEUF wasdone according to the above understanding and theresults in Table 2a,b. However, the 10 000 and 3000MWCO hollow fibers, without considering the permeateflux and operation time, were used in the first and secondstep, respectively, and the initial CPC concentration wascontrolled at 4% in each step. The solution sample in eachstep was without any adjustment. The pH of each samplewas closed to 7.0. After MF, only little thuringiensin inthe supernatant was lost (Table 3). Table 3 shows thatthe thuringiensin recoveries for the first, second, andfinal retentate in each step were 67.2%, 24.0%, and91.2%, respectively. The recovery for the final retentatecould be further improved to a higher value by using thelower MWCO fiber in the second step. The proposedscale-up two-step MEUF process operated at room tem-perature for thuringiensin recovery was definitely simple,economic, and effective. This two-step MEUF processcould be used to recover thuringiensin from the super-natant of the B. thuringiensis fermentation broth.

Fly Control Bioassay. Thuringiensin is a secondarymetabolite of microorganisms that is known to be effec-tive as a means of controlling the larval development of

flies. Up to now, fly control bioassay (Ignoffo and Gard,1970) has been the most commonly used method ofmonitoring the ability of thuringiensin to prevent pupaeof housefly larvae from developing into normal adults.Such analysis took 10 days to complete a single assay.An agar-based diet and procedure according to Ignoffoand Gard (1970) were used to assay the activity ofthuringiensin in the presence of CPC. Activity wasevaluated as larval mortality. The results were shownin Table 4. Larval mortality increased with increasingthuringiensin concentration. This result was in closeagreement with that of Mohd-Salleh and Lewis (1982).Most larval mortality occurred with the first few daysduring or shortly after molting. At low-concentrationtreatments, the spray-dried powder of thuringiensin withCPC further yielded shorter lethal times than the filteredsupernatant without CPC addition (Table 4), but the highconcentration of thuringiensin resulted in short lethal

Table 1. Effect of Micelle Formation Temperature onThuringiensin Recovery in MEUF

thuringiensin recovery (%)

micelle formationtemperature (°C)

4% w/v initialCPC concn

8% w/v initialCPC concn

(a) UF Membranes of MWCO 30 000Used in the MEUF Process

25 70.2 78.145 72.1 80.150 72.4 81.455 70.7 79.160 69.5 75.9

(b) UF Membranes of MWCO 10 000Used in the MEUF Process

25 76.1 79.845 76.2 81.950 76.7 83.555 76.5 81.760 73.4 79.8

Table 2. Effect of CPC Concentration on ThuringiensinRecovery (%) Conducted in Two-Step MEUF Process

stepUF membrane

(MWCO)initial CPCconcn (%)

thuringiensinrecovery (%)

(a) UF Membranes of MWCO 10 000Used in the Second Step of MEUF

first 30 000 4 70.28 78.1

second 10 000 1 81.7 88.82 84.4 90.43 87.1 91.74 88.6 92.9

(b) UF Membranes of MWCO 3000Used in the Second Step of MEUF

first 30 000 4 70.28 78.1

second 3000 1 84.9 90.92 87.2 91.83 89.6 93.24 91.2 94.6

Figure 6. Flowsheet of the two-step MEUF process consistingof microfiltration and two-step ultrafiltration.

584 Biotechnol. Prog., 1999, Vol. 15, No. 3

times. Finally, there were one or two deformed adultsshown up in the bioassay of low concentrations (e2.5ppm) of thuringiensin treatments after 10 days of expo-sure. This result was due to the response of feedingdeterrent effect of B. thuringiensis toxin to inhibit housefly pupae for developing into morphologically normaladults (Gould et al., 1991; Mohd-Salleh and Lewis, 1982).

Figure 7 is showing that larval mortality also increasedwith increasing thuringiensin concentration and that thespray-dried powder of thuringiensin with CPC undertreatment concentrations of 1.25 ppm or less seemedmore toxic to fly larvae than standard and filteredsupernatant of thuringiensin without CPC after 5 daysof incubation. Maybe there were some other toxic materi-als such as CPC surfactant that existed in the powder toenhance the effect of activity. Some surfactants have theeffect of dermal toxicity on insects (Morgan et al., 1988).The activity of CPC surfactant on fly larval mortality wasalso studied. CPC became more toxic at a concentrationof 500 ppm or higher in the diets after 5 days of exposure

(data not shown). Generally, the CPC concentration inthe spray-dried powder was about 30-fold of thuring-iensin used, that is, when a 20 ppm of thuringiensin wasused in an assay a 600 ppm of CPC was also added in atreatment. In comparison with the filtered supernatantand thuringiensin standard without any surfactant, thespray-dried powder with CPC surfactant had a goodenhanced effect of thuringiensin activity at 1.25 ppm orlower treatments (Figure 7). The enhanced effect of CPCsurfactant on bioassay results was very obvious. In thefuture, a cationic CPC surfactant will be an importantrole for thuringiensin in biopesticide formulation pro-cesses.

ConclusionThe addition of CPC in MEUF processes resulted in a

significant improvement in thuringiensin recovery. Therecovery increased with the initial CPC concentration inthe retentate up to a saturation limit. The recoveries ofthuringiensin/CPC micelles in MEUF experiments couldbe remarkably decreased by the addition of the inert salt(NaCl) for higher ionic strength. No significant changein the value of the recovery was observed in the MEUFprocess conducted at a high pH range of 6.8-9.0. Therecovery increased with increasing the initial CPC con-centration up to a saturation limit and was higher whena low MWCO membrane was used. The proposed two-step MEUF process achieved a recovery as high as 94.6%when an initial CPC concentration of 4% and a UFmembrane of MWCO 3000 were used in the second stepof the process, while the recovery was 91.2% for the scale-up process. The two-step MEUF process operated at roomtemperature for thuringiensin recovery was simple,economic, and effective, while the trouble in controllingmicelle formation temperature was avoided. This re-search demonstrated that the CPC concentration and theuse of low MWCO membranes are far more importantthan controlling the micelle formation temperature, interms of thuringiensin recovery. In bioassay study, it wasfound that the presence of CPC in the thuringiensinpowder enhanced the insecticidal activity, which maypartially be due to the fact that CPC is also toxic to flylarvae. The use of the cationic surfactant CPC not onlyincreased thuringiensin recovery in the MEUF processbut also contributed significantly in the activity ofthuringiensin against fly larvae. The CPC surfactant

Table 3. Effect of Two-Step MEUF Process onThuringiensin Recovery (%) (Process Steps Refer toFigure 6)

composition

vol ofsolution

(mL)

thuring-iensin

(g)CPC

(% w/va)

recoveries(% of startingthuringiensin)

starting supernatant 1000 2.50 0.0 100filtered supernatant 1000 2.40 0.0 96.0first step MEUF

process1000 2.40 4.0 96.0

first retentate 200 1.68 19.0 67.2first permeate 800 0.72 0.25 28.8second step MEUF

process800 0.72 4.25 28.8

second retentate 160 0.60 20.0 24.0second permeate 640 0.12 0.31 4.8two-step MEUF

processfinal retentateb 360 2.28 19.4 91.2final permeate 640 0.12 0.31 4.8

a % w/v ) CPC (g)/volume of solution (mL) x 100%. b Finalretentate was obtained from the second retentate mixed with thefirst retentate.

Table 4. Effect of Various Levels of Filtered Supernatantand Spray-Dried Powder of Thuringiensin on Mortality(%) at Larval Stage and Number of Deformed Adults at10 daysa

avb % mortality at larval stagedose of

thuringiensin(ppm) 1 day 2 days 3 days 4 days 5 days

avb no. ofdeformedadults at10 days

blankc 0.0 0.0 0.0 0.0 0.0 0.0A (20) 7.5 15.0 97.5 100.0 100.0 0.0B (20) 2.5 22.5 77.5 97.5 100.0 0.0A (15) 12.5 22.5 75.0 97.5 100.0 0.0B (15) 7.5 12.5 50.0 97.5 97.5 0.0A (5) 2.5 17.5 42.5 65.0 97.5 0.0B (5) 2.5 7.5 35.0 97.5 100.0 0.0A (2.5) 5.0 12.5 25.0 57.5 90.0 0.0B (2.5) 0.0 5.0 32.5 67.5 87.5 1.0A (1.25) 0.0 5.0 12.5 50.0 62.5 1.0B (1.25) 7.5 10.0 20.0 57.5 77.5 0.0A (0.625) 0.0 5.0 7.5 22.5 35.0 2.0B (0.625) 0.0 2.5 27.5 45.0 62.5 1.0a A: Filtered supernatant of thuringiensin without CPC addi-

tion obtained from the fermentation broth. B: Spray-dried powderof thuringiensin with CPC addition obtained from the two-stepMEUF process (the initial CPC concentration was 4% w/v in eachstep for MEUF). b Average of four replicates; 10 larvae/replicateper dose. c Blank indicates diets without any thuringiensin addi-tion.

Figure 7. Effect of different levels of standard, filteredsupernatant, and spray-dried powder of thuringiensin on larvalmortality (%) at 5 days. *Thuringiensin standard was suppliedby Dr. H. de Barjac, Institute Pasteur, Paris, France.

Biotechnol. Prog., 1999, Vol. 15, No. 3 585

could perform an important role for facilitating thuring-iensin in biopesticide formulations and applications inthe future.

AcknowledgmentThis work was supported under Research Grant NSC-

87-2622-E-259-002 from the National Science Council ofTaiwan, ROC. The support is greatly appreciated.

References and NotesBenz, G. On the Chemical Nature of Heat Stable Toxin of

Bacillus thuringiensis Berliner in Locusta migration. J.Invertebr. Pathol. 1966, 6, 381-383.

Bhat, S. N.; Smith, G. A.; Tucker, E. E.; Christian, S. D.;Scamehorn, J. F.; Smith, W. Solubilization of Cresols by1-Hexadecylpyridinium Chloride Micelles and Removal ofCresols from Aqueous Streams by Micellar-Enhanced Ultra-filtration. Ind. Eng. Chem. Res. 1987, 26, 1217-1222.

Campbell, D. P.; Dieball, D. E.; Brackete, J. M. Rapid HPLCAssay for Beta-Exotoxin of Bacillus thuringiensis. J. Agric.Food Chem. 1987, 35, 156-158.

Christian, S. D.; Bhat, S. N.; Tucker, E. E.; Scamehorn, J. F.;El-Sayed, D. A. Micellar-Enhanced Ultrafiltration of Chro-mate Anion from Aqueous Streams. AIChE J. 1988, 34, 2,189-194.

de Barjac, H.; Burgerjon, A.; Bonnefoi, A. The Production ofHeat-Stable Toxin by Nine Serotypes of Bacillus thuringien-sis. J. Invertebr. Pathol. 1966, 4, 537-538.

Dunn, R. O., Jr.; Scamehorn, J. F.; Christian, S. D. Use ofMicellar-Enhanced Ultrafiltration to Remove Dissolved Or-ganic from Aqueous Streams. Sep. Sci. Technol. 1985, 20, 4,257-284.

Dunn, R. O., Jr.; Scamehorn, J. F.; Christian, S. D. Concentra-tion Polarization Effects in the Use of Micellar-EnhancedUltrafiltration to Remove Dissolved Organic Pollutants fromWastewater. Sep. Sci. Technol. 1987, 22 (2 & 3), 763-789.

Gibbs, L. L.; Scamehorn, J. F.; Christian, S. D. Removal ofn-Alcohols from Aqueous Streams Using Micellar-EnhancedUltrafiltration. J. Membr. Sci. 1987, 30, 67-74.

Gould, F.; Anderson, A.; Landis, D.; Melaert, H. V. FeedingBehavior and Growth of Heliothis virescens Larvae on DietsContaining Bacillus thuringiensis Formulations or Endotox-ins. Entomol. Exp. Appl. 1991, 58, 199-210.

Hayashi, S.; Ikeda, S. Micelle Size and Shape of Sodium DodecylSulfate in Concentrated NaCl Solutions. J. Phys. Chem. 1980,84, 744-751.

Hsu, T. H.; Tzeng, Y. M.; Wu, M. M. Insecticidal Activity andHPLC Correlation of Thuringiensin from Fermentation andTwo-Phase Aqueous Separation Process. Pestic. Sci. 1997, 50,35-41.

Ignoffo, C. M.; Gard, I. Use of an Agar-Base Diet and HouseFly Larvae to Assay â-Exotoxin Activity of Bacillus thuring-iensis. J. Econ. Entomol. 1970, 63 (6), 1987-1989.

Ignoffo, C. M.; Gregory, B. Effect of Bacillus thuringiensisâ-Exotoxin on Larval Maturation, Adult Longevity, Fecun-dity, and Egg Viability in Several Species of Lepidoptera.Environ. Entomol. 1972, 1, 3, 269-272.

Issid, I.; Lavergne, M.; Lemordant, D. Using Micelles toDetermine the Pore Size of UF Membranes. J. Membr. Sci.1992, 74, 297-287.

Jong, J. Z.; Hsiun, D. Y.; Wu, W. T.; Tzeng, Y. M. Fed-BatchCulture of Bacillus thuringiensis for Thuringiensin Produc-tion in a Tower Type Bioreactor. Biotechnol. Bioeng. 1995,48 (3), 207-213.

Kandori, K.; Schechter, R. S. Selection of Surfactants forMicellar-Enhanced Ultrafiltration. Sep. Sci. Technol. 1990,25 (1 & 2), 83-108.

Kim, Y. T.; Huang, H. T. The â-Exotoxin of Bacillus thuring-iensis. 1. Isolation and Characterization. J. Invertebr. Pathol.1970, 15, 100-108.

Kreig, A.; Langenbruch, G. A. In Microbial Control of Pests andPlant Diseases 1970-1980; Burgess, H. D., Ed.; Academic:New York, 1981; pp 837-896.

Lecadet, M. M.; de Barjac, H. In Pathogenesis of InvertebrateMicrobial Diseases; Davidson, E. W., Ed.; Allanheld & Os-mum: Totowa, NJ, 1981; pp 293-321.

Levinson, B. L.; Kasyan, K. J.; Chin, S. S. Identification of Beta-Exotoxin Production Plasmids Encoding Beta-Exotoxin, anda New Exotoxin in Bacillus thuringiensis by Using HighPerformance Liquid Chromatograph. J. Bacteriol. 1990, 172,3172-3179.

Liu, C. M.; Tzeng, Y. M. Quantitative Analysis of Thuringiensinby High-Performance Liquid Chromatograph Using Adenos-ine Monophosphate as an Internal Standard. J. Chromatogr.Sci. 1998, 36, 7, 340-344.

Markels, J. H.; Lynn, S.; Radke, C. J. Micellar Ultrafiltrationin an Unstirred Batch Cell at Constant Flux. J. Membr. Sci.1994, 86, 241-261.

Markels, J. H.; Lynn, S.; Radke, C. J. Design of Micellar-Enhanced Ultrafilters. Ind. Eng. Chem. Res. 1995a, 34,2436-2449.

Markels, J. H.; Lynn, S.; Radke, C. J. Cross-Flow Ultrafiltrationof Micellar Surfactant Solutions. AIChE J. 1995b, 41, 9,2058-2066.

Missel, P. J.; Mazer, N. A.; Benedek, G. B.; Young, C. Y.; Carey,M. C. Thermodynamic Analysis of the Growth of SodiumDodecyl Sulfate Micelles. J. Phys. Chem. 1980, 84, 1044-1057.

Mohd-Salleh, M. B.; Lewis, L. C. Feeding Deterrent Responseof Corn Insects to â-Exotoxin of Bacillus thuringiensis. J.Invertebr. Pathol. 1982, 39, 323-328.

Morgan, R. L.; Rodson, M.; Scher, H. B. In Pesticide Formula-tion: Innovations and Applications; Cross, B., Scher, H. B.,Eds.; American Chemical Society: Washington, DC, 1988; pp131-141.

Os, N. M. V.; Daane, G. J.; Haandrikman, G. The Effect ofChemical Structure upon Thermodynamics of Micellizationof Model Alkylarenesulfonate. J. Colloid Interface Sci. 1991,141, 1, 199-217.

Pramauro, E.; Bianco, A.; Barni, E.; Viscardi, G.; Hinze, W. L.Preconcentration and Removal of Iron (III) from AqueousMedia Using Micellar-Enhanced Ultrafiltration. Colloid Surf.1992, 63, 291-300.

Pramauro, E.; Prevot, A. B.; Savarino, P.; Viscardi, G.; de laGuardia, M.; Cardells, E. P. Preconcentration of AnilineDerivatives from Aqueous Solutions Using Micellar-EnhancedUltrafiltration. Analyst 1993, 118, 23-27.

Sebesta, K.; Farkas, J.; Horska, K. In Microbial Control of Pestsand Plant Diseases 1970-1980; Burges, H. D., Ed.; Aca-demic: New York, 1981; pp 249-281.

Tsun, H. Y.; Tzeng, Y. M. In Proceedings 1994 Taipei-KyushuJoint Symposium on Chemical Engineering and Symposiumon Transport Phenomena and Application; Lu, W. M., Wu,W. T., Hsu, J. P., Koto, M., Nakashio, F., Hatate, Y., Eds.;National Taiwan University: Taipei, 1994; pp 403-408.

Tzeng, Y. M.; Chang, Y. N. In Proceedings: 7th Congress of AsianPacific Confederation of Chemical Engineer; Lin, O. C. C., Lee,J., Liou, C. T., Eds.; 7th APCChE Congress: Taipei, 1996;Vol. 2, pp 584-588.

Tzeng, Y. M.; Hsu, T. H. In Better Living through InnovativeBiochemical Engineering; Teo, W. K., Yap, M. G. S., Oh, S.K. W., Eds.; National University of Singapore: Singapore,1994; pp 595-597.

Tzeng, Y. M.; Young, Y. H. Penicillin-G Enhanced Productionof Thuringiensin by Bacillus thuringiensis sp. Darmstadien-sis. Biotechnol. Prog. 1995, 11, 231-234.

Tzeng, Y. M.; Young, Y. H. Production of Thuringiensin fromBacillus thuringiensis Using a Net-Draft-Tube ModifiedAirlift Reactor. World J. Microbiol. Biotechnol. 1996, 12, 1,32-37.

Wu, M. M.; Tzeng, Y. M.; Hsu, T. H. A Study on High-YieldFermentation of Thuringiensin Formation Monitored byHPLC Method. J. Technol. 1993, 8, 223-230.

Accepted April 1, 1999.

BP990053O

586 Biotechnol. Prog., 1999, Vol. 15, No. 3