generations. In this context, Price et al. (2003) indicated the need to use physical,
metabolic, and physiological constraints when modeling in silico adaptive
experiments. Along such line of thought, Ibarra et al. (2002) grew an E. coli K-12
strain for 700 generations on glycerol as a single carbon source. They observed
that the evolved population increased their growth rate from a suboptimal to near
optimal level as calculated previously by in silico. This result proved the usefulness of constrained metabolic models in the search for opportunities to improve
strains using adaptive evolution. Furthermore, Dykhuizen and Dean (1990)
showed that the relative fitness of two strains of E. coli growing in a mixed culture
with lactose as carbon source was explained by differences in two kinetic steps:
lactose permease and lactase. They emphasized the importance of finding the
metabolic bottlenecks to optimize adaptive evolution experiments.
Therefore, it seems quite possible to improve microbial strains, after few hundred
generations of adaptive selection, if the objective function (the phenotype) is related
to the growth process. For example, if certain enzyme activities are related to the
intake and utilization of a given carbon source. The problem is how to put constraints
forcing the evolving organism to improve specific functions. For example, related to
the improvement of a small set of enzymes. This is a promising approach because it
is leading toward the identification of few mutation sites related to strategic points of
the metabolic network, instead of searching throughout the complete genetic network. Such a new approach could be the way to combine random mutation during
adaptive evolution, with optimal in silico design of culture conditions.
11.5.6 The Use of Image Analysis for Automated Screens
of Microbial Strains
An interesting possibility for scoring superior mutants is the use of automated image
analysis of small colonies grown on agar plates (Loera and Viniegra-González 1998)
or directly on wheat bran particles (Couri et al. 2006, Dutra et al. 2008).
Loera and Viniegra-González (1998) used image analysis to estimate the growth
rate and pectinase potency of a collection of A. niger mutants. They found that
potency was inversely correlated to growth rate, in high or low levels of water
activities. This result can be related to the observation of Wösten et al. (1991) that
glucoamylase secretion is located at the tips of leading hyphae of A. niger. Apparently, enzyme secretion and fungal growth compete for the same space at the hyphal
tips. As a consequence, more potent strains would have lower hyphal extension rate.
Couri et al. (2006) and Dutra et al. (2008) developed an automated procedure of
image analysis for direct measurement of fungal biomass growing on the surface of
wheat bran particles. They found a linear correlation between the enzyme activities
and mycelial surface area. This is an important result because it provides a nondestructive method to measure biomass grown on irregular solid particles that are
intended to be used for actual SSF processes. A possible future development is the
use of HTS with image analysis using small samples of solid substrates together
11 New Horizons for the Production of Industrial Enzymes
331
metabolic, and physiological constraints when modeling in silico adaptive
experiments. Along such line of thought, Ibarra et al. (2002) grew an E. coli K-12
strain for 700 generations on glycerol as a single carbon source. They observed
that the evolved population increased their growth rate from a suboptimal to near
optimal level as calculated previously by in silico. This result proved the usefulness of constrained metabolic models in the search for opportunities to improve
strains using adaptive evolution. Furthermore, Dykhuizen and Dean (1990)
showed that the relative fitness of two strains of E. coli growing in a mixed culture
with lactose as carbon source was explained by differences in two kinetic steps:
lactose permease and lactase. They emphasized the importance of finding the
metabolic bottlenecks to optimize adaptive evolution experiments.
Therefore, it seems quite possible to improve microbial strains, after few hundred
generations of adaptive selection, if the objective function (the phenotype) is related
to the growth process. For example, if certain enzyme activities are related to the
intake and utilization of a given carbon source. The problem is how to put constraints
forcing the evolving organism to improve specific functions. For example, related to
the improvement of a small set of enzymes. This is a promising approach because it
is leading toward the identification of few mutation sites related to strategic points of
the metabolic network, instead of searching throughout the complete genetic network. Such a new approach could be the way to combine random mutation during
adaptive evolution, with optimal in silico design of culture conditions.
11.5.6 The Use of Image Analysis for Automated Screens
of Microbial Strains
An interesting possibility for scoring superior mutants is the use of automated image
analysis of small colonies grown on agar plates (Loera and Viniegra-González 1998)
or directly on wheat bran particles (Couri et al. 2006, Dutra et al. 2008).
Loera and Viniegra-González (1998) used image analysis to estimate the growth
rate and pectinase potency of a collection of A. niger mutants. They found that
potency was inversely correlated to growth rate, in high or low levels of water
activities. This result can be related to the observation of Wösten et al. (1991) that
glucoamylase secretion is located at the tips of leading hyphae of A. niger. Apparently, enzyme secretion and fungal growth compete for the same space at the hyphal
tips. As a consequence, more potent strains would have lower hyphal extension rate.
Couri et al. (2006) and Dutra et al. (2008) developed an automated procedure of
image analysis for direct measurement of fungal biomass growing on the surface of
wheat bran particles. They found a linear correlation between the enzyme activities
and mycelial surface area. This is an important result because it provides a nondestructive method to measure biomass grown on irregular solid particles that are
intended to be used for actual SSF processes. A possible future development is the
use of HTS with image analysis using small samples of solid substrates together
11 New Horizons for the Production of Industrial Enzymes
331