You are on page 1of 7

Generalization of Monod Kinetics

for Analysis of Growth Data with


Substrate Inhibition

J. H. T. Luong
Biotechnology Research Institute, National Research Council of Canada,
Montreal, Quebec, Canada, H4C 2K3

Accepted for publication February 7, 1986

The inhibitory effect of butanol on yeast growth has been PmS


studied for the strain Candida utilis ATCC 8205 growing
aerobically on butanol under batch conditions. A math- '= (Ks + S)(1 + S/Ki)
ematical expression was then proposed to fit the kinetic Equation (1) was used by Haldane' in 1930 to de-
pattern of butanol inhibition on the specific growth rate:
scribe enzyme inhibition by the formation of an inac-
tive complex of the enzyme with two substrate mol-
P=*[l-;]"
ecules. Boon and Laudelout3 later found that equation
The maximum allowable butanol concentration above
(1) was applicable for fitting their experimental results
which cells do not grow was predicted to be 9.16 g/L.The with Nitrobacter. The Haldane equation has then been
proposed model appears to accurately represent the ex- concluded by several researchers to provide an ade-
perimental data obtained in this study and the literature quate fit to a plot of p vs. S for growth at high substrate
data developed for a variety of batch culture systems at levels. As Ki approaches infinity, equation (1) ap-
widely ranging substrate concentrations.
proaches the familiar rectangular hyperbola of the
Monod relationship.
INTRODUCTION Edwards4 in 1970 pointed out that many other in-
hibitory models could be borrowed from enzyme ki-
Microbial growth can be inhibited in several different netics to fit kinetic data taken from the literature:
ways. The presence of inhibitors such as phenol, chlor-
ine, benzenoid, antibiotics, etc., in the growing me-
dium inhibits growth and may distort the metabolism
of microorganisms. Such interesting behavior plays a FmS
very important role in chemotherapy, and it has led to cL=
Ks + S + (S2/Ki)(l + S / K ) (3)
an enormous amount of research in theoretical and
applied biology. The other inhibitory mechanisms are E d ~ a r d showever,
,~ also found that equation (1) was
recognized as substrate and product inhibitions where no less consistently suitable than equation (2) or (3)
the substrate and/or product concentrations are al- since very large values were obtained for the fourth
lowed to increase to levels such that the growth de- parameter ( K ) in equations (2) and (3).
creases and ultimately ceases. The analysis of growth curves according to equation
Several mathematical models have been developed (1) has some complications when considering plots of
for quantifying the inhibitory effect of substrate on p vs. S because this model implies that the cells are
growth rate, and they are generally adaptations of capable of growing indefinitely. This is not what is
equations for substrate inhibition of enzymatic reac- observed in reality since there is a definite substrate
tions. concentration limit above which growth will cease
(complete inhibition).

Haldane Kinetics Linear Kinetics


Andrews' proposed that the effect of substrate con- Hinshelwood' studied the reduction of the growth
centration on specific growth rate could be governed rate of Bacterium Iactis perogenes by alcohols and
by the equation proposed the following relationship:

Biotechnology and Bioengineering, Vol. XXIX, Pp. 242-248 (1987)


0 1987 John Wiley & Sons, Inc. CCC 0006-3592/87/020242-07$04.00
= pm (e-SlK, - e-SlKs) (1 1)
p = p, - aP (4)
where P is product concentration and a is a constant. Again, equation (11) fails to predict the total inhi-
Equation (4)has also been employed by several in- bition concentration S, the level at which growth can-
vestigators for assessing product inhibition in the al- not occur, since p only approaches zero when S ap-
coholic fermentati~n.~.' proaches infinity.
Based on equation (4), Tseng and W a y ~ n a npro- ~,~ In this study a kinetic model has been proposed to
posed the following relations for correlating the growth quantify the effect of substrate inhibition on specific
data of Candida utilis, C . lipolytica, Arthrobacter AK growth rates. The model takes into account the be-
19, and Pseudomonas methanica: havior that substrates will act as inhibitors at higher
concentrations and behave as activators at lower lev-
els. The kinetic data developed for the growth of C.
utilis on n-butanol and the literature data have been
used to assess the applicability of the proposed model.
PmS
i(S - S*) when S > S* (6)
p=K,+s-
PROPOSED KINETIC MODEL
Here S* is a threshold substrate concentration below
which the organism grows apparently without inhibi- Levenspie112 recently proposed a generalized non-
tion. The discontinuity is a major drawback of this linear equation to account for the influence of ethanol
model. Literature data also indicate that the relation- production on the rate of alcoholic fermentation:
ship between p and S is not always a linear
p = E.L,s (1 - -E)n
Ks + S
Exponential Kinetics If common mechanisms are assumed, relations pro-
Aiba et al.IOproposed the following equation for cor- posed to describe product inhibition may suggest ad-
relating product inhibition data from the alcoholic fer- ditional empirical relationships to correlate substrate
mentation: inhibition:

pms e-PIKi
p==
(7)

Edwards4 suggested that relations proposed to de- where S, is the maximum substrate concentration above
scribe product inhibition may be borrowed to correlate which growth is completely inhibited.
substrate inhibition: As illustrated in Figure 1, at low substrate concen-
trations the specific growth rate increases with increas-
ing substrate concentration in accordance with the
Monod equation. Equation (13) will have a maximum
value when dplds = 0, and the value p* can be cal-
Like equation (I), equation (8) fails to predict the culated as the value of p corresponding to a substrate
maximum substrate concentration at which growth will concentration S*:
be completely inhibited. It is worth noting that when
SIK, << 1, equation (8) becomes equivalent to equa-
tion ( I ) , and both these equations approach equation
(9) by a Taylor series analysis:

(9)
The value of p then decreases as substrate concen-
tration is further increased and the magnitude of con-
Teissier Kinetics stant n will indicate the type of relation between p
and S .
The assumption of diffusion-controlled substrate Linear. When n approaches unity, equation (13) be-
supply leads to equation (lo), which was originally comes equivalent to equation (9) and has a maximum
derived by Teissier": value corresponding to the following substrate con-
p = p,(l - e-s/Ks) centration:
(10)
Combining this mechanism with a protective diffu- S* = Ks [(l + SmlKs)1/2- 13 (15)
sional limitation of high and inhibitory concentrations, Nonlinear (concavity upward). A rapid initial drop
Edwards4proposed the following model for correlating in the growth rate followed by a slow decrease to zero
the growth data: occurs when n > 1.

LUONG: GENERALIZATION OF MONOD KINETICS 243


ture. The dissolved oxygen concentration was main-
tained close to the saturation level by sparging air through
the bioreactor at 1 L/min (IVVM). The exit air stream
was passed through a reflux cooler in which butanol
vapor was condensed and returned to the fermentor,
thus minimizing the evaporation of butanol during the
course of the fermentation.

Measurement of Cell Growth


Growth was followed with a Klett-Summerson pho-
toelectric colorimeter at 420 nm. A relationship be-
tween optical density and dry weight was established
a priori, and it was assumed that butanol did not alter
the optical properties of the cells in suspension appre-
ciably.

0
..I2 t 1, ;2 j3 ;4 rs :6
I

,
Assay of n-Butanol
Butanol was determined by using a gas-liquid chro-
matography model AGC 311 (Carle Instruments, CA).
sIS, A nickel column, 183 cm long and 0.318 cm outside
diameter, packed with 80/100 mesh Chromosorb 101
Figure 1. p/p, plotted against S/S, as a function of n. (Johns-Manville, CO) was used.

RESULTS AND DISCUSSION


Nonlinear (concavity downward). A slow initial de-
crease in the growth rate followed by a rapid decrease Values of exponential specific growth rate p were
to zero occurs when n < 1 . determined for each initial butanol concentration So by
It is obvious that the proposed kinetic model is of plotting In X vs. time (figure not shown). These data
the generalized Monod type and accounts for both sub- were then fitted to equations (l), (8), ( 1 l), and (13) by
strate availability and substrate inhibition. nonlinear least-squares regression techniques for es-
timation of the biokinetic constants. It should be noted
that the least-squares method is sensitive to the guessed
MATERIALS AND METHODS
values of the biokinetic constants, and it is quite pos-
sible to obtain numerical values with no physical mean-
Microorganism
ing. Such a problem was partly overcome by allowing
Candida utilis ATCC 8205 was used in this study. the computer routine to search for a minimum least
squares only with a bounded domain of allowable val-
ues for the constants (Table I). An upper limit for the
Culture Medium
constant K , value was set to be 1 g/L. It is worth noting
The medium used had the following c o m p o ~ i t i o n ' ~ that the K , value obtained for microbial growth on
(for 1 L medium): 5 g (NH4)Z SO4, 0.5 g MgS04-7H20, carbon sources generally ranges from 1 to 50 mg/LI4.
3.5 g Na2HP04, 3 g KH2P04, 0.4 g yeast extract, 10 However, recent studies on the growth C. boidinii, a
mL 0.4% Fe EDTA solution (ferric salt of ethylene-di-
aminetetraacetic acid). n-Butanol, reagent grade, was
filter sterilized and aseptically added to the fermentor Table I. Bounded domain of allowable values for the biokinetic
before inoculation. The same medium was used for the constants.a
inoculum preparation.
Equations ( I ) , (S), and (11) K, 2 K,
PZ Pm 5 3PL:
Fermentor
Equation (13)
Batch experiments were carried out in a 2-L fer- K, s ig / ~
mentor with a 1-L working volume. During fermen- s, 2 s;
tation the pH and the liquid temperature were auto- a pt is the highest experimentally observed specific growth rate.
matically controlled at 4.5 and 30°C respectively- Mild
9 S; is the highest experimentally observed initial substrate concen-
agitation was provided to maintain a homogeneous cul- tration at which growth still occurs.

244 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 29, FEBRUARY 1987


Table 11. Parameter values, substrate inhibition models for the growth of C. utilis ATCC 8205 on 1-butanol.

Proposed model, eq. (13) 0.694 0.554 0.130 - 9.16 1.03 0.89 I
Equation (1)= 1.2% 0.570 0.515 2 - 1.01 -
Equation (8)b 1.224 0.575 0.694 3.96 - 1.35 -
Equation (1 1 0.701 0.555 0.290 5.07 - 0.88 -

a S* = (K,K,)In
S* = 4 K , {[I + 4 (K,IK,)]'n - I}

methanol-utilizing yeast, have shown a K , value as high predict the inhibitory effect of substrate concentration
as 0.64 g/L15. on the specific growth rate. The model also predicted
The values for the model parameters p,, K,, K i , S,, that the specific growth rate p increased with increas-
and n have been estimated and summarized in Table ing butanol concentrations from 0 to 1.03 g/L in ac-
11. The comparison of the proposed inhibition model cordance with the Monod equation. However, butanol
and the literature model with the experimental data concentrations above 1.03 g/L were inhibitory since
was illustrated in Figure 2. As shown in this figure, at the value of p decreased with a further increase in the
butanol concentrations of up to 5.75 g/L, both equa- initial butanol concentration. The parameter p has a
tions (1) and (8) represented the experimental data rea- maximum value of 0.554 h- corresponding to 1.03 g/L
sonably well. However, at higher concentrations equa- initial butanol concentration. In general, such values
tion (1) was no Ionger applicable for correlating the agreed quite well with those predicted by equations
growth data vs. initial substrate concentrations. Figure (l), (8), and (1 1) (Table 11).
2 also indicates that equation (1 1) was inferior to equa- The best-fitted value for S, the maximum allowable
tions ( 1 ) and (8) at low substrate concentrations (<8 butanol concentration above which cells do not grow,
g/L) but reflected the trend in the data slightly better was estimated to be 9.16 g/L. It is worth noting that
at higher concentrations. butanol is somewhat toxic to microorganisms, and the
The kinetic data obtained in this study were ex- maximum tolerance level is about 10 g/L8.
tremely well represented by the proposed model to Comparison of the goodness of fit of these equations

0 1 I
\?\
0 1 2 3 4 5 6 7 8 9
s (9iL)
Figure 2. Inhibitory effect of butanol or yeast growth: (0)Experimental data (S, p) [dL,
h-'I, (0.8,0.55), (1.6,0.54), (3,0.468), (4,0.402), (5.7,0.285), (8.2,0.106); ( - - - - - - - - - - ) eq.
(8); (0) eq. (1); (.) eq. (11); (-) proposed model.

LUONG: GENERALIZATION OF MONOD KINETICS 245


Table 111. Goodness of fit of the inhibitory kinetic models.” ent from equation (8) at both the 0.05 and 0.01 prob-
ability levels.
x (pi - pi)* df 02 = C(pi - bi)*/df
The experimental values of p vs. S reported in the
Proposed model 0.0000025 2 0.00000125 literature have also been used to evaluate the appli-
Equation ( 1 ) 0.01632 3 0.00544 cability of the proposed model. As shown in Figure 3,
Equation (8) 0.004391 3 0.00146 equation (13) represented the literature very well. From
Equation (1 1) 0.01446 3 0.00482
Tseng and Wayman’s data,8 the proposed model pre-
a Abbreviations: df = degree of freedom = number of data - dicted that the maximum ethyl acetate concentration
number of fitting parameters; p i = observed specific growth rate; above which C. lipolytica ATCC 8661 does not grow
pi = predicted specific growth rate. was 50 g/L (Table IV). The value predicted by Tseng
and Waymans by means of equation (6) was 42.1 g/L.
A plot between p vs. S derived from the growth data
to the experimental data was made using the F-test for of C. utilis on sodium acetate4 definitely indicated that
equality of variances in accordance with the methods the curve obtained did not conform to the straight-line
employed by Edwards4 (Table 111). The 02 value ob- relationship postulated by Tseng and Wayman.8 Again,
tained from the proposed model was lowest while that equation (13) was found to represent this set of ex-
of equation (1) was highest. Equation (1 1) offered little perimental data very well. The values of S , and n were
improvement in the fit to the data compared to equation predicted to be 39.22 g/L and 0.446, respectively.
(l), and equation (8) was the second best. It was, there- KortanI6 studied the growth of Arfhrobacter AK 19
fore, decided to perform the F-test for equality of var- on n-butanol and observed that the initial growth rate
iances between equation (8) and the proposed model increased and then decreased with an increasing con-
[eq. ( 1 3 1 . centration of n-butanol. The author also attempted to
The value of F was calculated as fit the experimental data to three of Edward’s functions
equations (1)-(3) and obtained a fair agreement. Based
F =
a2 = 1168
- (16) on this kinetic data, the maximum butanoi concentra-
0% tion above which cells do not grow was predicted to
where dLis the larger variance and 0% is the smaller be 10 g/L. The applicability of the proposed model for
variance. representing the experimental data of Kortan is illus-
F tables tabulated for 0.05 and 0.01 probability levels trated in Figure 2 and Table IV where the d value was
with a df(numerator) of 3 and df(denominator) of 2 very small (0.00048), indicating a very good fit.
indicate values of 19.2 and 99.2, respectively. Since Asthana” studied substrate inhibition of P . meth-
the calculated F value (1168) was very much larger m i c a by methanol and observed that equation (I) was
than either one of the tabulated values, it may be con- only applicable for fitting the experimental data for
cluded that the proposed model is significantly differ- concentrations of methanol up to 30 mL/L. As illus-

-
1

U
P
.4.

I
0

Figure 3. Inhibitory effect of substrates on growth rates (data reported in the literature):
(0) C. lipolytica on ethyl acetate*; (0)C. utilis on sodium acetate4; (0)P.methanica on
methanol1’; ( +) Arthrobacter AK-19 on n-butanolB6;(-) proposed model.

246 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 29, FEBRUARY 1987


Table IV. Model parameters for the literature data.

C . utilis on sodium acetate" 0.41 0.8% 39.22 0.446 0.00226


Arthrobacter on n-butanolb 0.335 0.113 10 1.653 0.000048
C.lipolytica on ethyl acetate' 0.335 0.7% 50 1.41 O.oooO5487
Pseudomonas methanica on methanold 0.206 0.478 x 42.43 0.835 O.oooO107
Methylomonas mucosa on methanol' 0.913 6.4 20.74 0.776 0.000425

From ref. 4.
From ref. 16.
From ref. 8.
From ref. 17.
From ref. 18.

trated in Figure 3, equation (13) was found to represent lizing bacteria reported in the literature. l9 The other
the data of Asthana quite well. The proposed model kinetic constant, K , in the Monod model, was found
also predicts that the maximum methanol concentra- to be 6.4 g/L methanol. This value is two orders of
tion above which cells do not grow was 42.43 g/L. It magnitude higher than the value of 0.12 g/L reported
is interesting to observe that the value of K , was very for Hansenula polymorpha, a thermophilic methanol-
small (0.478 x lo-' g/L). Such an observation thus utilizing yeast whose growth kinetics also fit the Monod
indicated that the relationship between p and S at The study on the growth of C. boidinii, an-
methanol concentrations higher than 1.58 g/L could other methanol-utilizing yeast, shows a K , value as
simply be represented by the equation high as 0.64 g/L.Is However, no literature values of K ,
for methanol-assimilating bacteria are available for
comparison. Tam and Finn18also observed that meth-
anol concentrations above 1% v/v were inhibitory and
that the Monod kinetics was no longer applicable for
where the values of Sm and n were determined to be representing the growth rate data.
42.43 g/L and 0.835, respectively. The applicability of equation (3) for fitting the ex-
Tam and FinnI8 studied the growth of Methylomonas perimental data of Tam and Finn'* is illustrated in Fig-
mucosa NRRL B-56% on methanol and observed that ure 4, where an upper limit for the constant K , value
at methanol concentrations less than 1% v/v the spe- was set to be 6.4 g/L. The proposed model predicts
cific growth rate data fits a Monod model for substrate- that the maximum tolerable methanol concentration
limited growth. The maximum specific growth rate was was 20.74 g/L and that the small value of d obtained
determined to be 0.725 h-l, i.e., about 3 times higher (0.00042) indicates a good fit. In general, methanol is
than the average value for most of the methanol-uti- a toxic substrate for bacteria; even for methanol-

0 DATA

- E O l 131

0.2

0.1

0 1

0 5 10 I5 20
s C9/LI
Figure 4. Inhibitory effect of methanol on growth rates of Methylomonas mucosa.'8

LUONG: GENERALIZATION OF MONOD KINETICS 247


utilizing organisms a concentration below 1% v/v may 6. I. Holzberg, R. K. Finn, and K. H. Steinkraus, Biotechnol.
inhibit the growth of many strains.'* Bioeng., 9, 413 (1967).
Undoubtedly, the proposed kinetic model appears 7. T. K. Ghose and R. D. Tyagi, Biotechnol. Bioeng., 21, 1401
( 1979).
to be useful for representing the kinetics of substrate 8. M. C. Tseng and M. Wayman, Can. J. Microbiol., 21,994 (1975).
inhibition. The model is of the generalized Monod type 9. M. Wayman and M. C. Tseng, Biotechnol. Bioeng., 28, 383
and accounts for both substrate stimulation at low con- (I 976).
centrations and substrate inhibition at high concentra- 10. S. Aiba, M. Shoda, and M. Nagatani, Biotechnol. Bioeng., 10,
tions. The model has the capability to predict values 845 (1968).
11. G. Teissier, Rev. Sci., No. 3208 (Extract), 209 (1942).
of S,, the maximum substrate concentration above 12. 0. Levenspiel, Biotechnol. Bioeng., 22, 1671 (1980).
which growth is completely inhibited. 13. A. C. Kormendy and M. Wayman, Can. J . Microbiol., 20, 225
(1 974).
14. S. J. Pirt, Principles ofMicrobe and Cell Cultivation (Blackwell,
Oxford, England, 1975).
References 15. P. Pilot and A. Prokop, Biotechnol. Bioeng., 17, 1717 (1975).
16. K. Kortan, M.A.Sc. Thesis, University of Toronto, Ontario,
1. J. F. Andrews, Biotechnol. Bioeng., 10,707 (1%8). Canada, 1972.
2. J. B. S. Haldane, Enzymes (Longmans, Green, New York, 1930, 17. H. N. Asthana, Ph.D. Thesis, University of Pennsylvania, Pitts-
and MIT Press, Cambridge, MA, 1965). burgh, PA, 1972.
3. B. Boon and H. Laudelout, Biochem. J . , 85, 440 (1962). 18. K. T. Tam and R. K. Finn, ACS Symp. Ser., 45, 58 (1977).
4. V. H. Edwards, Biotechnol. Bioeng., 27, 679 (1970). 19. K. T. Tam, Ph.D. Thesis, Cornell University, Ithaca, NY, 1975.
5. C. N. Hinshelwood, The Chemical Kinetics of the Bacterial 20. P. W. Levine and C. L. Cooney, Appl. Microbiol., 26, 982
Cells (Oxford University Press, London, 1946). (1973).

248 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 29, FEBRUARY 1987

You might also like