Framework for online optimization of recombinant protein expression in high-cell-density Escherichia coli cultures using GFP-fusion monitoring

A framework for the online optimization of protein induction using green fluorescent protein (GFP)-monitoring technology was developed for high-cell-density cultivation of Escherichia coli. A simple and unstructured mathematical model was developed that described well the dynamics of cloned chloramphenicol acetyltransferase (CAT) production in E. coli JM105 was developed. A sequential quadratic programming (SQP) optimization algorithm was used to estimate model parameter values and to solve optimal open-loop control problems for piecewise control of inducer feed rates that maximize productivity. The optimal inducer feeding profile for an arabinose induction system was different from that of an isopropyl-beta-D-thiogalactopyranoside (IPTG) induction system. Also, model-based online parameter estimation and online optimization algorithms were developed to determine optimal inducer feeding rates for eventual use of a feedback signal from a GFP fluorescence probe (direct product monitoring with 95-minute time delay). Because the numerical algorithms required minimal processing time, the potential for product-based and model-based online optimal control methodology can be realized.

Biocatalytic transformation of [(2-Hydroxyethyl)thio]acetic acid and thiodiglycolic acid from thiodiglycol by Alcaligenes xylosoxydans ssp. xylosoxydans (SH91)

A Gram-negative bacterium, Alcaligenes xylosoxydans ssp. xylosoxydans (SH91), consumed thiodiglycol (TDG), the nontoxic hydrolysis product of sulfur mustard, as a primary carbon source and transformed TDG to commercially relevant chemical precursors, [(2-hydroxyethyl)thio]acetic acid (HETA) and thiodiglycolic acid (TDGA). Aerobic fed batch and repeated batch experiments were run to compare the molar yields of HETA and TDGA that result under different operating policies. In repeated batch experiments, 35% of the TDG was converted to HETA. Under the conventional batch process and a repeated fed batch process, the HETA yields were reduced (21% and 18%, respectively), while the yield of TDGA was increased (47% and 31%,respectively). This work demonstrated that cell growth associated biocatalytic transformations were manipulated to achieve a desired byproducts profile through an understanding of the specific reaction and cell growth kinetics and by altering the reaction operating policy accordingly.

Monitoring GFP-operon fusion protein expression during high cell density cultivation of Escherichia coli using an on-line optical sensor

Synthesis of an operon fusion protein was investigated in batch and fed-batch cultures at high cell densities of recombinant Escherichia coli JM105 [pBAD-GFP::CAT]. Glucose-limited growth was achieved without accumulation of inhibitory byproducts allowing high cell densities (110 g L(-1) DCW) to be attained. This was believed to be the highest reported value for dry cell mass of E. coli strain JM105 expressing two recombinant proteins. Transcription of the two reporter genes, green fluorescent protein (GFP) and chloramphenicol acetyltransferase (CAT), was under the control of the p(BAD) promoter of the araBAD (arabinose) operon. Each protein was independently translated via separate ribosome binding sites. CAT served as a model recombinant protein product to illustrate the noninvasive quantitative reporting ability of GFP during high cell density fermentations. Expression of GFP was monitored on-line using an intensity-based optical sensor. A linear correlation between the on-line GFP intensity and the enzymatic activity of CAT allowed for in vivo real-time quantitative monitoring of a fermentation product under conditions of high biomass concentration and high productivity.

Observations of metabolite formation and variable yield in thiodiglycol biodegradation process : impact on reactor design

The complete microbial degradation of thiodiglycol (TDG), the primary hydrolysis product of sulfur mustard, by Alcaligenes xylosoxydans ssp. xylosoxydans (SH91) was accomplished in laboratory-scale stirredtank reactors. An Andrews substrate inhibition model was used to describe the cell growth. The yield factor was not constant, but a relationship with initial substrate concentration has been developed. Using a substrate-inhibition and variable-yield kinetic model, we can describe the cell growth and substrate consumption in batch and repeated batch fermentations. Several reactor-operating modes successfully degrade TDG concentration to below 0.5 g/L. According to the experimental results, the two-stage repeated batch operation has the best degradation efficiency, and it also can degrade 500 mM TDG (= 60 g/L) to 5 mM (= 0.7 g/L) in <5 d. A hypothesis for explaining variable-yield and byproduct formation based on the capacity and utilization of metabolic loads is presented.

Bioreactor strategies for the treatment of growth-inhibitory waste: an analysis of thiodiglycol degradation, the main hydrolysis product of sulfur mustard

The microbial degradation of thiodiglycol, the primary hydrolysis product of sulfur mustard, by a pure culture of Alcaligenes xylosoxydans ssp. xylosoxydans (SH91) was accomplished in laboratory scale stirred tank reactors. This is a major component of the overall biodegradation process proposed for the complete mineralization of sulfur mustard. Several configurations were evaluated for degradation efficiency including batch, repeated batch, continuous stirred tank reactor (CSTR), and two-stage series CSTR. The repeated batch reactor provided the highest degradation rate of thiodiglycol. Further, this method degraded thiodiglycol in the liquid broth to below the detection limits (0.03 mM). Both batch and repeated batch experiments were simulated by an unstructured mathematical model. Simulation results were in agreement with the experimental data, particularly at low TDG concentration (around 30 mM). This study demonstrates the degradation of thiodiglycol using bioreactors and, more generally, is an experimental study of bioreactor designs for the degradation of growth-inhibitory substances.

Reactor comparisons for the biodegradation of thiodiglycol, a product of mustard gas hydrolysis

An environmentally benign method for the mineralization of sulfur mustard has been proposed involving chemical hydrolysis of sulfur mustard to thiodiglycol, and then the biological degradation of thiodiglycol to generate biomass and gaseous carbon dioxide. Alcaligenes xylosoxidans (SH91) was isolated based on its ability to utilize thiodiglycol as a sole carbon source. This article compares different biological reactor designs and experimentally assesses their relative effectiveness in degrading thiodiglycol using pure cultures of SH91. The reactor configurations studied are batch, continuous stirred-tank reactor (CSTR), and CSTR with cell recycle. From the results, it is clear that the CSTR with cell recycle offers superior performance for a given residence time or volume. These pure culture data are necessary for accurate design of a pilot-scale system where mixed cultures will be employed because of a possible incomplete chemical hydrolysis step.

Optimization for a recombinant E. coli fed-batch fermentation

The operating strategy that produces the maximum foreign protein expression for a fed-batch process is desired. This is achieved by using a feasible quadratic programming (FSQP) algorithm with a structured model that describes cell growth and product formation for recombinant E. coli. Optimization calculations for a fed-batch culture have not been performed with a model of this complexity up to this point. A constraint on the maximum cell concentration was included. For a fixed value of batch time, the results show that the optimal time profile of feed flow rate can increase the yield of foreign protein by 12-29% over a constant feed rate policy. Also, it was found that the computation time for the FSQP algorithm can be reduced significantly by considering suboptimal profiles of the feed rate, with a minor effect on calculated protein yield.

The effect of cellular energetics on foreign protein production

Escherichia coli strain F-122 was used to determine if there are additional physiological effects, other than decreasing energetic efficiency accompanied by the excretion of the acetate, on foreign protein production. This organism was the host for expressing HIV582-beta-galactosidase fusion protein under the control of the trp promoter, with ampicillin resistance. By comparing parallel batch cultures with and without acetate addition, it was found that the presence of acetate in the media did not influence beta-galactosidase activity. In these experiments, it appears that the low protein productivity often observed during acetate formation is the result of inefficient cell metabolism, rather than acetate acting as a specific inhibitor of protein production.

On the mechanism of growth of cells (Bacillus amyloliquefaciens) in the mixed aqueous two-phase system

The growth of Bacillus amyloliquefaciens in the aqueous two-phase system, made up of polyethylene glycol, dextran, and water, was investigated. Generally, Bacillus partitions in the dextran phase, but the magnitude of the separation depends largely on the overall composition of polymers in the phase system. The kinetics of growth of Bacillus amyloliquefaciens was studied in the polyethylene glycol-rich continuous phase, dextran-rich dispersed phase, and in the mixed phase. From the kinetic data it appears that increasing the overall polymer composition causes the cells to adsorp at the interface. On the other hand, partition measurements indicate that increasing polymer concentrations make the cell partitioning more one-sided. This anomaly is explained by studying the interfacial adsorption of cells via dynamic surface tension measurements.

Effects of recombinant plasmid size on cellular processes in Escherichia coli

The effects of recombinant plasmid size on cell growth and viability, plasmid copy number, and synthesis of plasmid-encoded protein were investigated in Escherichia coli using plasmid pUC8 and four recombinant derivatives containing inserts of Drosophila melanogaster DNA of 1.7-6.0 kb. Growth in log phase was unaffected by plasmid size, but as plasmid size increased, maximum cell density decreased and, with the largest plasmid, cell death was accelerated after the stationary phase was reached. There was also a correlation between increasing plasmid size and decreased viability at high ampicillin concentrations, resistance to which is conferred by the plasmids. These effects were shown not to be due to transcription or translation of Drosophila sequences carried on the recombinant plasmids. Cells harboring the largest plasmid, pBS5 (8.7 kb), fared poorly in competition with plasmid-free cells in mixed cultures, compared with cells harboring pUC8 (2.7 kb). In addition, pBS5 was harbored at significantly fewer copies per cell than pUC8 at all phases of growth and supported much less production of the plasmid-encoded protein, beta-lactamase, than did pUC8. The results suggest that recombinant plasmid size may be an important parameter in the optimization of large-scale production of plasmid-encoded proteins.

Pharmacokinetics of chlorpromazine-induced miotic response in rabbits

The time variation of changes in the chlorpromazine-induced pupil diameter decrease was studied following varying bolus and slowly infused intravenous doses administered to rabbits. The observed pharmacological response data were coverted, via, the use of a dose-effect curve, to values theoretically corresponding to relative biophasic drug levels. These values were, in turn, used to construct a linear pharmacokinetic model of the drug bioavailability input equilibrium pharmacological response output dynamics of the system. The use of a time domain, MULTIFIT, computerized method of fitting the data to obtain a pharmacokinetic model was compared to the use of a frequency response, PLTEST, approach. The fidelity of the model in quantitatively relating the time course of systemic drug bioavailability to observed pupil response was verified by the satisfactory agreement obtained by directly comparing experimentally known amounts of drug intravenously infused with corresponding values computed from observed changes in pupil size. The applicability of using pharmacological data for quantitative bioavailability and pharmacokinetic analysis of chlorpromazine is demonstrated. This finding is particularly significant because no suitable chemical or radiological direct assay technique exists for determining levels of chlorpromazine, except for high doses, in body fluids.

Bioavailability and pharmacokinetic analysis of chlorpromazine-induced rectal temperature depression in rabbits

Chlorpromazine-induced depression of rectal temperatures in rabbits kept at 20 degrees was used to determine relative biophasic drug levels corresponding to observed hypothermic response intensities, which then served to establish a triexponential linear mathematical model describing drug transference and drug action in this system. Comparisons of various experimentally known, slow intravenous infusion drug inputs of chlorpromazine with drug inputs computed by deconvolution, using the derived model and observed temperature depressions, served to verify the accuracy of the model for the 0.50-4.0 mg/kg dosage range.

Optimal Control of Linear Distributed Parameter Systems With Constrained Inputs

The optimal open loop control of systems described by a set of linear partial differential equations is investigated. The performance index is of quadratic type and the mean square error is considered as a special case. Energy type inequality constraints are imposed on the control inputs. The problem is formulated as a minimization problem in Hilbert space. The necessary and sufficient conditions for a minimum are obtained and it is proved that these conditions yield the global minimum. It is shown how the solution to the constrained problem can be obtained from the solution of the unconstrained problem. The optimal control functions satisfy Fredholm integral equations with symmetric kernels. The paper presents an example where the solution is obtained by eigenfunction expansion.