Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
, Laura Varas, Jorge Vald es, Cecilia Cabello, Lester Rodrguez & Manuel Mansur
Center for Genetic Engineering and Biotechnology (CIGB), Biotechnology Development Unit, Fermentation De-
velopment Department, P.O. Box 6162, Havana 10600, Havana City, Cuba
EA
Q
EA
, (1)
d(X V )
dt
=
FB
X V, (2)
d(M V)
dt
= M
F
F
FB
X V
Y
XM
, (3)
d(P V)
dt
=
dC
MPI
dt
=
FB
X V
Y
xp
, (4)
where t : time (h);
F
,
C
: fresh media and culture
densities (g ml
1
);
EA
,
WA
: densities of exhausted
and wet air (g ml
1
); F: ow of feeding fresh medium
(l h
1
); Q
EA
, Q
WA
: volumetric ow of exhausted and
wet air (ml h
1
); X: cell concentration (g dry wt l
1
);
V: culture volume (l);
FB
specic growth rate at fed-
batch phase (h
1
); M, M
F
: methanol concentration in
253
Table 1. Correlations and coefcients of determination (R-squared values) obtained from the fermentation
data.
Correlation CD
a
Correlation CD
a
V = 4.0 0.01279 t R
2
= 0.7998 X V = 148.29 e
0.0926t
i
R
2
= 0.9874
1
Y
XM
= 0.9801 +
0.0261
R
2
= 0.8299 X V = 149.22 e
0.0114t
i
R
2
= 0.9471
V
S
V
= 1.0121 0.001 X
W
R
2
= 0.8126 C
MPI
= 0.9132 X V 138.77 R
2
= 0.8650
X = 0.3388 X
W
R
2
= 0.9205 F = 0.048 X V
a
Coefcient of determinations.
the culture and feeding medium (g l
1
); P: concen-
tration of MPI (mg MPI l
1
); C
MPI
: total quantity
of MPI (mg); Y
XM
biomass/methanol yield (g dry
wt g
1
); Y
XP
: biomass/MPI yield (g dry wt mg
1
).
The inuence of the last two elements of the right
side of Equation (1) can be neglected in well-designed
bioreactors with an adequate heat transfer area in the
top condenser that greatly reduces the water loss from
fermentation media. Nevertheless, in the case under
study, this difference is approximately equal to the
water loss in the culture. Finally, assuming almost con-
stant the culture density throughout the fermentation,
the expression (1) can be rewritten as:
C
dV
dt
=
F
F
H
2
O
Q
Lost
, (5)
where:
H
2
0
: density of water lost during fermentation
(g ml
1
); Q
Lost
: average volumetric owof water loss
during fermentation (l h
1
).
From Equation (3) must be demonstrated that
methanol-feeding rate is equal to:
F =
FB
X V
M
F
Y
XM
. (6)
Equation (6) can be rearranged and equalized ac-
cording to Pirts equation (Pirt 1975) or the relation
between 1/Y
XM
and 1/. Thus, the new equation is:
1
Y
XM
=
FB
X V
M
F
F
=
1
Y
T
XM
+
m
M
, (7)
where: m
M
: maintenance coefcient (g g dry wt
1
h
1
); Y
T
XM
: true Biomass/methanol yield (g dry
wt g
1
).
Differential Equations (2), (4) and (5) must be
rewritten in an integral form, and the following equa-
tions are obtained:
C
MPI
=
1
Y
XP
X V
1
Y
XP
(X V)
0
, (8)
X V = (X V)
o
e
FB
(t t
0
)
= (X V)
0
e
FB
t
i
,
(9)
V = V
0
H
2
O
0
Q
Lost
dt
0
F dt
, (10)
where (X V)
0
: total biomass obtained in batch phase
(g dry wt); t
0
: duration of batch phase (h); t
i
: induction
time (h).
Finally, the relation between the culture and super-
natant volumes is:
V
S
V
=
C
X
W
S
, (11)
where V s: supernatant volume (l);
s
: supernatant
density (g ml
1
); X
W
: cell density (g wet wt. l
1
).
Results and discussion
The cell mass at the end of the batch phase was ap-
prox. 38 g dry wt l
1
and it was reached after 25 h.
After the initial glycerol was consumed, a methanol
concentration of 1% (v/v) was suddenly established in
the culture medium. After adaptation to the new C-
source, the pH controller was positioned in a new set
point value (4.6 0.1). At this moment, a continuous
methanol feeding (M
F
= 775.8 g l
1
, supplemented
with 12 ml PTM
1
l
1
), with periodic increment in the
ow was initiated. Every 12 h, a new feed ow rate
was xed after calculating an average value, substitut-
ing in Equation (6) the values of 0.012-h
1
and 0.4-g
dry wt g
1
for the desired specic growth rate and
biomass/methanol yield, respectively.
Applying the above explained mass balance equa-
tions and the experimental fermentation data, a set of
254
Fig. 1. Experimental data of four 7 l fermentations (B.E. Marubishi, Tokyo, Japan) performed under the same conditions as described in
Materials and methods (the symbols , , , correspond to fermentation from 1 to 4) and moldel prediction (solid lines) are obtained from
the application of the unstructured model. The dotted lines represent the 95%-condence intervals in each model. (A) Variation of culture
volume in four fermentations, (B) relation between de l/Y
XM
and 1/
FB
in the induction phase, for fermentations 1 and 2, (C) experimental
relations between dry and wet well wt. concentrations () and between the supernatant and culture volumes ratio (V s/V) and wet cell weight
concentration () for the four fermentations performed, (D) kinetics for total biomass at different induction times for fermentations 1 and 2,
(E) kinetics for total MPI in the supernatant at different induction times (for fermentation 1 only), (F) a typical pH prole and constant ow
steps xed for all fermentations during the fed-batch phase of each fermentation process (bold solid lines).
correlations was obtained, which correspond to the
searched unstructured model (Table 1). In Figure 1
can be observed the experimental data and tted the
achieved correlations, observing a good correspon-
dence between them. Other authors have employed
this kind of models for simulation of the fermenta-
tion process at large-scales and to optimize the process
(Zhang et al. 2000a, DAnjou & Daugulis 2001).
From the determined unstructured model, some
fermentation parameters could be obtained, which are
presented in Table 2.
The specic growth rate and biomass/methanol
yield used to calculate the feed ow in Equation (6)
are in good correspondence with the same parameters
shown in Table 2 obtained from the dry cell concen-
tration time chart (Figure 1D) and the Pirts relation
(Figure 1B). The specic growth rate is typical for
the Mut
S
phenotypes as reported by Brierley et al.
(1990). A value of biomass/glycerol yield higher than
the value reported by Roels (1983) for other methy-
lotrophic species was obtained, maybe due to the
incomplete consumption of glycerol in the inocula. On
the other hand, the biomass/methanol yield and main-
255
Table 2. Fermentation parameters obtained from the achieved
unstructured model.
Fermentation parameter Value
a
Specic growth rate
Batch phase
B
= 0.093 0.013 h
1
Fed-batch phase
FB
= 0.011 0.002 h
1
Biomass substrate yields
On glycerol Y
XG
= 0.619 0.08 g dry wt g
1
On methanol Y
XM
= 0.306 0.06 g dry wt g
1
Y
T
XM
= 1.02 0.56 g dry wt g
1
Biomass product yield Y
XP
= 1.09 0.24 g dry wt mg
1
Maintenance coefcient m
M
= 0.026 0.012 g dry wt
1
h
1
Densities
Culture
C
= 1.01 0.06 g ml
1
Supernatant
S
= 1.00 0.01 g ml
1
a
Average values standard deviation.
tenance coefcient were similar to those obtained by
Zhang et al. (2000b) for a Mut
+
(Methanol Utilization
Plus) GS115 strain expressing intracellularly a heavy-
chain fragment C of botulinum neurotoxin, serotype
A. The expression levels obtained were signicantly
lower than the values obtained by Wang et al. (2001),
reaching up-to 1.5 g MPI l
1
in the culture super-
natant, but higher than results reported by Kjeldsen
et al. (1999).
Conclusions
An unstructured model was developed for a Mut
S
P. pastoris strain expressing the MPI. The model de-
scribes the relationships between the total biomass
grown under a specic growth rate, both in batch and
in fed-batch phases. The average values of specic
growth rates of 0.093 and 0.011 h
1
for the batch
and fed-batch phases respectively, was obtained. The
calculated biomass/substrate yields showed that there
is a little overestimate of this variable in the batch
phase, but in the fed-batch phase, the results are in
clear correspondence with values reported by other
authors. In the analyzed case the water loss during
the fermentation is signicant and for this reason the
variation of the volume along the fermentation intro-
duces changes in the traditional presentation of mass
balance equations. However, robustness of unstruc-
tured model based on mass balance equations permits
the accurate representation of the experimental data
values. Studies are in progress to test this model for
other secreted MPI constructions in P. pastoris.
References
Brierley RA, Bussineau C, Kosson R, Melton A, Siegel RS (1990)
Fermentation development of recombinant Pichia pastoris ex-
pressing the heterologous gene: bovine lysozyme. Ann. N.Y.
Acad. Sci. 589: 350362.
Cregg JM, Cereghino JL, Shi J, Higgins DR (2000). Recombinant
protein expression in Pichia pastoris. Mol. Biotechnol. 16: 23
52.
DAnjou MC, Daugulis AJ (2000) Mixed-feed exponential feed-
ing for fed-batch culture of recombinant methylotrophic yeast.
Biotechnol. Lett. 22: 341346.
Goeddel DV, Kleid DG, Bolivar F, Heyneker HL, Yansura DG
(1979) Expression in Escherichia coli of chemically synthe-
sized genes for human insulin. Proc. Natl. Acad. Sci. USA 76:
106110.
Kjeldsen T, Pattersson AF, Hach M (1999) Secretory expression and
characterization of insulin in Pichia pastoris. Biotechnol. Appl.
Biochem. 29: 7986.
Markussen J, Damgaard U, Diers I, Fiil N, Hansen MT, Larsen P,
Norris F, Norris K, Schou O, Snel L (1986) In: Theodoropoulos
D, ed. Peptides. Berlin: Walter de Gruyter & Co., pp. 189194.
Pirt SJ (1975) Principles of Microbe and Cell Cultivation. Oxford:
Blackwell Scientic Publications.
Roels JA (1983) Energetics and Kinetics in Biotechnology. Amster-
dam: Elsevier Biomedical Press.
Thim L, Hansen MT, Norris K, Hoegh I, Boel E, Forstrom J, Am-
merer G, Fiil NP (1986a) Secretion and processing of insulin
precursors in yeast. Proc. Natl. Acad. Sci. USA 83: 67666770.
Thim L, Norris K, Hansen MT (1986b) Insulin precursors, process
for their preparation and process for the preparation of human
insulin. Eur. Pat. Applic. 86302133.3.
Wang Y, Liang ZH, Zhang YS, Yao SY, Xu YG, Tang YH, Zhu
SQ, Cui DF, Feng YM (2001) Human insulin from a precur-
sor overexpressed in the methylotrophic yeast Pichia pastoris
and a simple procedure for purifying the expression product.
Biotechnol. Bioeng. 73: 7479.
Zhang W, Inam M, Meagher MM (2000a) Fermentation strategies
for recombinant protein expression in the methylotrophic yeast
Pichia pastoris. Biotech. Bioproc. Eng. 5: 275287.
Zhang W, Bevins MA, Plantz BA, Smith LA, Meagher MM (2000b)
Modeling Pichia pastoris growth on methanol and optimizing
the production of a recombinant protein, the heavy-chain frag-
ment Cof botulinum neurotoxin, serotype A. Biotechnol. Bioeng.
70: 18.