JOURNAL

OF FERMENTATION AND BIOENGINEERING

Vol. 83, No. 6, 561-564. 1997

Scale-Up of Bacillus thuringiensis Fermentation Based on

Oxygen Transfer
ELfAS RAZO FLORES,’ FERMfN Pl?REZ,Z AND MAYRA DE LA TORRE2*
Department of Environmental Technology, Wageningen Agricultural University, P.O. Box 8129, 6700 EV
Wageningen, The Netherlands’ and Department of Biotechnology, CZNVESTA V-ZPN,
A. P. 14-740, $7000 Mexico City, Mexico2
Received 5 July 19961Accepted17 April 1997

A correlation between spore productivity and kLa, was found for Bacillus thuringiensis var. kurstaki. Since
insecticidal crystal proteins synthesis depends on the highly oxygen-dependent sporulation process, the fer-
mentation process was scale-up on the basis of the product kLa p, where p is the fermentor total pressure. The
process was scale-up from a 14-1 fermentor to a 1,100-Z fermentor. Both reactors did not have the same
geometrical design. Even though the specilic growth rate decreased with the change of scale, the scale-up had
a positive effect, since the total fermentation time was decreased while the biomass yield and the sporulation
efficiency were unchanged. Therefore kLa p is a practical tool for scaling up B. thzuingiensis processes where
product synthesis depends on sporulation.

[Key words: Bacillus thuringiensis,scale-up, oxygen transfer, kLa, spore productivity]

Bioinsecticides based on Bacillus thuringiensis (Bt) appropriate criterion for scaling-up the process. In this
represent the highest volume of natural insecticides work kLa p, where kLa is the volumetric oxygen-transfer
produced worldwide and are used as part of integrated coefficient and p is the vessel total pressure, was used to
pest management system (1). Bt is the natural host for scale-up a Bt fermentation.
a variety of plasmid-encoded insecticidal crystal proteins
(ICPs) that are toxic to important coleopteran, lepidopte-
MATERIALS AND METHODS
ran, and dipteran insect pest. Most ICP gene transcrip-
tion is concomitant with sporulation (2). The crystal pro- Microorganism B. thuringiensis var. kurstaki HD-
teins are included in large crystalline inclusions called 73, was acquired through the USDA Insect Pathology
parasporal crystals (b-endotoxin) and can represent up Research Unit, Brownsville, Texas, USA. An active cul-
to 25% of the dry weight of the sporulated cells (3). ture was maintained by streak inoculating nutrient agar-
Currently, Bt is produced in submerged culture as a slants tubes, incubating them at 30°C for 2 d, and then
spore-crystal suspension that is later formulated to a storing at 4°C. The culture was transferred to new agar
commercial insecticide (4). slants every 6 months.
Several reports on Bt fermentation scale-up have been Culture media Nutrient broth (Bioxon) was used
published but none of them have considered criteria for inoculum growth and a complex medium described
generally used to scale-up fermentation processes, i.e. earlier was utilized for propagation (11).
mass-transfer coefficient #,a), power per unit of liquid, Fermentation conditions Bench scale fermentations
gas hold-up, Reynolds number, Peclet number, etc. were performed in a 14-1 total volume stirred tank fer-
Paige and Cooper (5) reported the scale-up of a fed- mentor (CFS-314, New Brunswick Scientific Co., USA).
batch culture of Bt to produce /3-exotoxin; they just Full-scale runs were carried out in a l ,lOO-1 total volume
limited the supply of the C-source during the non- locally-built-fermentor. The dimensions of both fermen-
growth-associated production phase. Gandman et al. (6) tors are summarized in Table 1. Temperature was con-
proposed that bioactivity thermograms could be applied trolled at 30°C. pH was controlled at 7.0 only during
to industrial scale-up fermentations, specifically to Bt the vegetative and transient phases of the growth cycle.
process, and Wecht-Lifshift et al. (7) reported the scale- Aeration rate and stirrer speed are mentioned for each
up of Bt var. berliner fermentation based on thermo- experiment. DO was monitored by a polarographic oxy-
gram data, obtained by calorimetric methods. The ther- gen electrode (Ingold Co., Switzerland). Concentrations
mogram allowed to define a precise harvesting time. of O2 and CO2 in the exit gas were measured with an
In the fermentation of Bt many authors have pointed Oxygen Analyzer Servomex 540 A (UK) and an Infrared
out that oxygen plays an important role in the sporu- CO2 Analyzer Servomex Ana Tek PSA 402 (UK), respec-
lation and d-endotoxin production. Dulmage (8) and tively.
Luthy and Ebersold (9) noted that growth and sporula- An inoculum growing in the exponential phase was
tion of Bt could be optimized using high aeration rates. obtained from a spore suspension through subcultures
Foda et al. (10) reported that cell counts, sporulation in nutrient broth as described earlier (11). A 2% (v/v)
titers and endotoxin potency of Bt var. entomocidus inoculum ratio was used. The fermentation ended when
varied with the level of aeration. 90% of the spores were released.
Since oxygen is one of the controlling parameters in Oxygen transfer measurements In all cases the
Bt fermentation, k,_a or a related parameter, might be an volumetric oxygen-transfer coefficient was determined by
the oxygen balance method. kLa was evaluated by sulfite
* Corresponding author. oxidation (0.2 M Na2S03 and 0.001 M CuS04.5Hz0),

561
562 FLORES ET AL. J. FERMENT. BIOENO.,

1
TABLE 1. Dimensions of bench and full scale fermentors
Fermentor Bench scale Full-scale
14 1100 350-
Total volume (r)
Working volume (f) 7 700
Six flat blade turbines 2 2 300-
Di (m) 0.07 0.27 ,^
4 (m) 0.21 0.80 k 250-
Dt/Di 3 3 -a
HI. b) 0.21 1.5 y” 200-
WDt 1 1.87
150-

whereas kra was determined during the fermentation

runs at the maximum oxygen transfer rate (12). The 14-1
fermentor was operated at a total pressure of 0.77 atm
100

501
100
-1 ,
200
,
300
,
400
, ,
500
. ,
600
1
(Mexico City’s atmospheric pressure) and a Cr* value of
0.187 mmol I-* (water at 30°C and 0.77 atm) was consi- k,a (h-l)
dered. For the 1,100~1fermentor the modified equation
FIG. 1. Relationship between the volumetric oxygen transfer
proposed by Votruba et al. (13) for large-scale reactors coefficients measured by sulfite oxidation &,a) and during the fer-
was used. This fermentor worked at a total pressure of mentation runs of E. thuringiensis var kurstaki (kLa’). Segmented
1.85 atm and a CL*=0.454mmol f-i (water at 30°C and lines represent 95% confidence limits. Symbols: B, bench-scale;
1.85 atm) was thus used. A, full-scale.
Analyses Total cell and spore counts were deter-
mined in a Neubauer chamber. Glucose was analyzed by
the dinitrosalysilic acid method (14). Biomass dry weight Full-scale studies kLa was evaluated at aeration
was calculated from the total cell count assuming a cell rates of 0.2 wm (V,=[Link]-I), 0.4 wm (V,=[Link] m
weight of 2.3 pg/bacteria (11). s-l), 0.6vvm (V,=0.015msP1) and 0.8wm (V,=O.O2m
s-l) and five stirrer speeds (1.67rps, 3.33 rps, 5.5rps,
6.67 rps and 8.33 rps). The relation between kLa, superfi-
RESULTS AND DISCUSSION
cial gas velocity and stirrer speed could be expressed as:
Bench scale studies kg was evaluated at aeration
kLa = 904 N.88 Vso.98 $=0.9967 (2)
rates of 0.8vvm (V,=0.0028ms-1) and [Link] (I’,=
0.0035 m s-r) and five stirrer speeds (6.67 rps, 8.33 rps, Scale-up The fermentation process was scaled-up
10 rps, 11.67 rps and 13.33 rps) using tap water. The stan- on the basis of kLa p, to correct for the driving force.
dard error of five determinations for each condition was In full-scale fermentors, the pressure increases greately,
less than 2%. Based on the experimental data obtained inducing inconsistencies when comparing the results of
at both aeration rates the relation between kra, super- large-scale fermentors with those from bench-scale ex-
ficial gas velocity (V,) and stirrer speed (N) could be periments. Aiba et al. (17) suggested that penicillin and
expressed as: streptomycin fermentations could be scaled up using this
basis. Bailley and Ollis (18) found a good correlation
kLa= 84.3 i@.49Vso.*O *=0.9991 (1) between kLa p and the vitamin Blz yield, although results
Since kLa values obtained by sulfite oxidation are for benchtop unit were generally below those obtained
often overestimated in regard to those observed in real for fermentation in larger vessels.
fermentations (15), a correlation between kLa and kLa’ At bench scale, the maximal spore volumetric produc-
was intended. To obtain this correlation, kLa’ was mea-
sured during fermentation runs at different stirrer speeds
(8.33rps, lOrps, 11.67rps and 13.33rps) and an aera-
tion rate of 1 wm (V,=O.O035 m s-r). A linear correla-
tion between kLa’ and kLa in the range of kLa 170 hP1
to 570 h-l was obtained (Fig. 1).
In the fermentation runs the period of oxygen limita-
tion during the transient growth phase and early sporula-
tion depended on the stirrer speed used (Fig. 2). The
final spore concentration (X,) and the volumetric spore
productivity (Pv) were a linear function of kLa for the
range tested (Fig. 3). Rowe and Margaritis (16) found
that Et var. kurstuki HD-1 has a maximal oxygen de-
mand during the transient growth phase when the
acetate accumulated during the vegetative growth is con-
sumed. Foda et al. (10) reported a marked increase in
viable cell and spore concentrations when the ratio total 25
0 5 10 15 20 3
volume: operation volume was decreased in shaken
flasks. These results confirm that oxygen is critical for Time (h)
sporulation. Since the parasporal crystal synthesis de- FIG. 2. Evolution of DO for B. thuringiensis var kurstaki at
pends on sporulation (2) it seems adequate to use kLa p different stirrer speeds at bench scale. Arrow length is proportional
as a scale-up criterion. to the period of oxygen limitation.
VOL. 83, 1997 SCALE-UP OF B. THURINGIENSIS 563

2.41
22-
2.0-
18-

1.6-

1.4-

12-

1.0-

1, ., , ., , ( ., , t 0.8 -
100 200 300 400 500 600 700

k,a (h-l) -I
50 100 150 200 250 300 350 400 450 500 550 600

FIG. 3. Final spore concentration (X,) and volumetric spore k,a p ( atm he’)
productivity (Pv) as function of k~a at bench scale. Symbols: (a) X,;
(a) P,. Linear regression for X,: X,=2.67+0.00364 kLa, R=0.99102; FIG. 4. Spore productivity of B. thuringiensis var kurstaki as
linear regression for P,: P,=O.51888+0.00245 kta, R=0.98418 function of kLa p. Symbols: E, bench-scale; n, full-scale. Linear
regression: P,=O.47442+0.00339 kLa p, R=0.92816

tivity (1.88 x 10” spores 1-l h-l) was reached at 13.3 rps
and 1 wm which corresponded to a kLa of 570 h-l and As Fig. 4 indicates, the correlation between spore
a kLa p of 440 atm h-l. The same value of kLa p was productivity and kLa p obtained from bench-scale experi-
obtained in the l,lOO-1 fermentor at 6.33 rps, 0.3 wm ments seems to be adequate for full-scale data, neverthe-
and a total pressure of 1.85 atm. Additional runs at full- less more experiments are needed. Our results suggest
scale were carried up at a kLa p of 500 atm h-l (5.83 rps that oxygen transport limits spore productivity in the
and 0.4 wm). The average results of full-scale and bench range tested and that the maximum P, has not been
runs are compared in Table 2. achieved.
At full-scale p and the time elapsed since exponential kLa’ values in full-scale runs were evaluated using the
growth started until 90% of spores released were smaller corresponding CL* and assuming CL neglegible (12). As
than those from bench-scale experiments. On the other shown in Fig. 1, these values were within the 95% confi-
hand Ytis was similar. Holmberg et al. (19) found that dence limits as calculated from bench scale experiments.
the ,u of Bt var. thuringiensis was reduced from 1.3-1.4 This suggest that the correlation between kLa’ and kLa
h-l to 0.95 h-l when they scaled-up from a 8-1 reactor to for the bench fermentor could fit for the full scale fer-
a 1,000-f. However, YXjswas slightly higher in the large- mentor in the range tested.
scale fermentor. Since Holmberg et al. studied the ex- The differences observed between the bench and full-
otoxin production and the transcription of this gene scale runs were presumably related to the mixing and
does not depend on sporulation, the effect of the scale- flow characteristics of each fermentor. The circulation
up on sporulation was not discerned. Unfortunately they time is generally not the same in reactors of different
studied the scale-up effects in a single fermentation run scale, especially if they do not have the same geometrical
and did not use a specific scale-up criterion. design. It is well known that when a scale-up is done
We are not aware of previous communications on the on a certain basis, it is necessary to be aware that other
effect of scale-up on sporulation of Bt. Our results sug- mixing and flow properties are different (17, 18).
gested a reduction in time needed for sporulation since The present work demonstrated that kLa p is applica-
the time elapsed since exponential growth started until ble for the scale-up of Bt processes where sporulation,
90% of spores released was shorter at large-scale i.e. 8-endotoxin production, is important. kLa can easily
although p was smaller. It is important to point out that be determined in any reactor and the large number of
the sporulation efficiency was not affected by the change empirical correlations available for kLa estimation (20)
of scale. could be employed to design a fermentor and fix the
operation conditions needed for a particular Bt produc-
TABLE 2. B. thuringiensis fermentation parameters at bench and tivity. Consequently the product kLa p is a practical tool
full scales for scaling up Bt processes.
Fermentor Bench-scale Full-scale
N (w) 13.33 6.33 5.83 NOMENCLATURE
wm (min-‘) 1 0.30 0.40
kLa (h- ‘) 570 238 270.70 C,* : oxygen concentration in bulk liquid phase in equili-
kLa’ (h- ‘) 293 117.56 118.66 brium with the oxygen partial pressure in bulk gas
kLa p (atm h-l) 440 500 phase, mmol IP1
X, (spores X 10~9/ml) 4.8 (0.3) 4Y(O.7) 5 (0.7) CL : oxygen concentration in bulk liquid phase, mmol I- I
P, (spores X lo-“//. h) 1.8 (0.2) 2.1 (0.1) 2.2 (0.1) Di : stirrer diameter, m
yx/, (-) 0.36 (0.03) 0.32 (0.03) 0.40 (0.03) Dt : tank diameter, m
0.78 (0.01) 0.68 (0.02) 0.67 (0.03)
22
HL : liquid height, m
/t”
T:hhTI’ 25.5 (2.5) 20 (1.5) (1.5) kLa : volumetric oxygen-transfer coefficient in liquid film
Values in parenthesis are STD. measured by sulfite oxidation, h -I
564 FLORES ET AL. J. [Link]..

kra’ : volumetric oxygen-transfer coefficient in liquid film tech’85 Europe, Geneve (1985).
measured during fermentation runs, h-l 7. Wecht-Llfshitz S.C., Gadman, M., and Zomer, E.: Process
N : stirrer speed, rps optimization and scale-up of the Bacillus thuringiensisfermenta-
tion. Isr. J. Entomol., 23, 239-246 (1989).
P : total pressure, atm
8. Dulmage , T. H.: Production of bacteria for biological control
Pv : volumetric spore productivity, spores I-’ h-l of insects, p. 129-139. In Papavizas, G. C. (ed.), Biological
tT : time elapsed since exponential growth started until control of crop production. Beltsville Symposia in Agricultural
90% of spores released, h Research. Allanhed, Osmun and Co., Totowa, N. J. (1981).
X, : final spore concentration, spores ml-’ 9. Luthy, P. and Ebersold, H. R.: Bacillus thuringiensis as a bac-
Yx/S : biomass yield, gbiomassper gducose terial insecticide: basic considerations and application, p. 35-
v, : superficial gas velocity, m ssl 74. In Kurstak, E. (ed.), Microbial and viral pesticides. Marcel
wm: air volume per liquid volume per minute, STP con- Dekker, New York (1982).
ditions 10. Foda, M. S., Sahuna, H. F., and Selim, M.: Factors affecting
growth physiology of Bacillus thuringiensis. Appl. Microbial.
In : specific growth rate, h-l Biotechnol., 22, 50-52 (1985).
11. Rodriguez, M. and de la Torre, M.: Effect of the dilution rate
ACKNOWLEDGMENTS on the biomass yield of Bacillus thuringiensis and determina-
tion of its rate coefficients under steady state conditions. Appl.
We are grateful to the IDRC-Canada and COSNET-Mexico for Microbial. Biotechnol., 45, 546-550 (1996).
financial supporting this research and to Orlando Melchy and Rey- 12. Kaeppeli, 0. and Fiechter, A.: On the metodology of oxygen
nold Farrera for their technical assistance. E. Razo-Flores also transfer coefficient measurements. Biotechnol. Lett., 3, 541-546
acknowledges the fellowship support of CONACYT-Mexico. (1981).
13. Votruba, J., Sobotka, M., and Prokop, A.: Evaluation of aera-
REFERENCES tion capacity from cultivation data files: application to large
scale fermentations. Biotechnol. Bioeng., 19, 1553-1556 (1977).
1. Feitelson, J. S., Payne, J., and Kim, L.: Bacillus thuringiensis, 14. Miller, G. L.: Use of dinitrosalicylic acid reagent for determina-
insects and beyond. Bio/Technol., 10, 271-275 (1992). tion of reducing sugar. Anal. Chem., 31, 426-428 (1959).
2. Malvar, T., Gawron-Burke, C., and Baum, A. J.: Overexpres- 15. Moresi, M. and Patete, M.: Prediction of KLa in conventional
sion of Bacillus thuringiensis HknA, a histidine protein kinase stirred fermentors. J. Chem. Technol. Biotechnol., 42, 197-210
homology, bypasses early spo-mutations that result in CryIIIA (1988).
overproduction. J. Bacterial., 176, 4742-4749 (1994). 16. Rowe, G. E. and Margarltls, A.: Endocellular fatty acid compo-
3. Agaisse, H. and Lereclus, D.: How does Bacillus thuringiensis sition during batch growth and sporulation of Bacillus thurin-
produce so much insecticidal crystal protein? J. Bacterial., 177, giensis kurstaki. J. Ferment. Bioeng., 77, 503-507 (1994).
6027-6032 (1995). 17. Aiba, S., Humphrey, E. A., and Millis, F. N.: Scale-up, p.
4. Rowe, G. E. and Margarltis, A.: Bioprocess development in 195-208. In Biochemical engineering, 2nd ed. Academic Press
the production of bioinsecticidas by Bacillus thuringiensis. Inc., New York (1973).
CRC Crit. Rev. Biotechnol., 6, 87-127 (1987). 18. Bailey, J. E. and OlIIs, D. F.: Scaling of mass transfer equip-
5. Paige, M. R. and Cooper, R.D.: Scale-up of beta-exotoxin ment, p. 508-512. In Biochemical engineering fundamentals,
production in fed-batch Bacillus thuringiensis fermentations, 2nd ed. McGraw-Hill., Int. Ed., Singapore (1986).
p. 146-149. In Proceedings of 5th European Congress on 19. Holmberg, A., Sievaneo, R., and Carlberg, G.: Fermentation
Biotechnology, Copenhagen, Denmark (1990). of Bacillus thuringiensis for exotoxin production: process analy-
6. Gandman, M., Vecht, S., Zomer, E., and Er-el, Z.: The use of sis study. Biotechnol Bioeng., 22, 1707-1724 (1980).
bioactivity monitor for process optimization and scale-up of 20. AtkInson, B. and Mavituna, F.: Gas-liquid mass transfer and
Bacillus thuringiensis fermentation, p. 665-670. In The world mixing, p. 712-751. In Biochemical Engineering and Biotech-
biotech report 1985, volume 1. Europe Proceedings of Bio- nology Handbook, 2nd ed. McMillan, Mexico (1991).