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.     

Potential of real-time measurement of GFP-fusion proteins

Building on the basic design concepts of Randers-Eichhorn [Biotechnol. Bioeng. 55 (1997) 921], an on-line, real-time robust, steam sterilisable optical sensor for monitoring green fluorescent protein (GFP) has been developed. A general cloning vector for fusion expression proteins was constructed, allowing expression of both GFP and the target protein as a fusion. Cultivations were carried out at the 20l scale with the signal from the sensor being relayed directly to the control system of the bioreactors. The production of GFP was then measured on-line, the signal was interfaced directly with other controlling parameters, thereby allowing the microbial process to be controlled directly based on recombinant protein expression. A positive expression correlation between on-line and off-line data was obtained. Protein accretion measured off-line was quantified using both LC-MS and plate reader assays. The potential of such a sensor for many aspects of process development is considerable and we have developed a working system which allows the optimisation of production conditions, for example, linking pH control directly to the fusion protein. Results are also presented that illustrate GFP does not alter the cultivation characteristics of the target protein when compared to the native construct. Whether GFP expressed as a fusion influences the solubility of the target protein is also discussed.     

Biofilm development in a membrane-aerated biofilm reactor: effect of intra-membrane oxygen pressure on performance

The effect on intra-membrane oxygen pressure at a constant carbon substrate loading rate on the development of biofilms of Vibrio natrigens in a membrane aerated biofilm reactor (MABR) was investigated experimentally and by mathematical modelling. A recently reported technique (Zhang et al., 1998. Biotechnol. Bioeng. 59: 80-89) for the in situ measurement of the substrate diffusion coefficients in a growing biofilm and the mass transfer coefficients in the boundary layer at the biofilm liquid interface was used. This aided the study of the effect of the heterogeneous biofilm structure and also improved the reliability of the model predictions. The different intra-membrane oxygen pressures used, 12.5, 25 and 50 kPa, with acetate as the carbon substrate, showed a marked effect on the initial biofilm growth rate, on acetate removal rate, particularly in thick biofilms and on biofilm structure. The model predicted the substrate limitation regimes, the location of the active biomass layer within the biofilms and the trends in oxygen uptake rate through the membrane into the biofilms. During the development of the biofilms, the biofilm thickness and the intra-membrane oxygen pressure were found to be the most important parameters influencing the MABR performance while the effect of biofilm structure was less marked.     

A new configuration of the Zeiss LSM 510 for simultaneous optical separation of green and red fluorescent protein pairs.

The power and simplicity of genetically encoded fluorophores (fluorescent proteins, FPs) have drawn many molecular biologists to light microscopy. First generation FPs suffered from overlapping excitation and emission spectra, which limited their use together in pairs (Patterson et al., J Cell Sci 2001;114 (Part 5):837-838). Image acquisition and processing techniques, collectively known as linear unmixing, have been developed to separate overlapping fluorescence signals encountered in the imaging of FP pairs and also in FRET. These specialized techniques are not without their potential drawbacks, including limitations on sensitivity and time-resolution for live cell imaging, and the risk of artifact in the hands of nonspecialists. With the advent of a new generation of red-shifted FPs (Shaner et al., Nat Biotechnol 2004;22:1567-1572; Verkhusha and Lukyanov, Nat Biotechnol 2004;22:289-296) careful selection of excitation sources and emission filters obviate the need for linear unmixing when simple two channel imaging of FPs is required. Here we introduce a new configuration of the Zeiss LSM 510 laser scanning confocal microscope, optimized for live cell imaging of green fluorescent protein (GFP) together with spectral variants such as mRFP1 and mCherry using standard photo-multipliers. A 2 mW, 594 nm HeNe laser was chosen as the excitation source for the red FP. This wavelength efficiently excites the aforementioned red variants without limiting the detection range of GFP emission during simultaneous two-channel imaging. Compared to excitation of GFP and mCherry at 488 and 543 nm, excitation at 488 and 594 nm approximately doubles the sensitivity of GFP detection and eliminates bleed-through of GFP into the mCherry channel. However, sensitivity of mCherry detection is decreased by 30%, suggesting the need for red FPs having longer emission peaks. Practical advantages to the simultaneous optical separation of FPs with nonoverlapping emission spectra include simplicity, robustness, reduced risk of artifact, and increased sensitivity during live cell imaging.     

Dynamic simulation of cyclic batch anaerobic digestion of cattle manure

Cyclic batch reactors with periodical feeds and extractions, are often used in cattle manure anaerobic digestion. The dynamic behavior of this type of reactor was simulated in this study. The kinetic model developed by I. Angelidaki et al. [Biotechnol. Bioeng. 42 (1993) 159], together with microbial growth kinetics, conventional material balances for an ideally cyclic batch reactor, liquid-gas interactions, and liquid phase equilibrium chemistry were used in this study. The model showed good agreement with the experimental data of R.I. Mackie and M.P. Bryant [Appl. Microbiol. Biotechnol. 43 (1995) 346], and R. Borja et al. [Chem. Eng. J. 54 (1994) B9]. The effects of hydraulic retention time (HRT), organic loading rate, reactant concentrations, feeding interval, and initial conditions such as pH and ammonia concentration on process performance can be evaluated by the dynamic model. Also simulation results show that the equilibrium conditions can be considered for CO2 distribution between liquid and gas phases, especially for processes with long retention times.     

Dual color microscopic imagery of cells expressing the green fluorescent protein and a red-shifted variant

The green fluorescent protein (GFP) from the jellyfish, Aequorea victoria, has become a versatile reporter for monitoring gene expression and protein localization in a variety of cells and organisms. GFP emits bright green light (λ_max = 510 nm) when excited with ultraviolet (UV) or blue light (λ_max = 395 nm, minor peak at 470 nm). The chromophore in GFP is intrinsic to the primary structure of the protein, and fluorescence from GFP does not require additional gene products, substrates or other factors. GFP fluorescence is stable, species-independent and can be monitored noninvasively using the techniques of fluorescence microscopy and flow cytometry [Chalfie et al., Science 263 (1994) 802–805; Stearns, Curr. Biol. 5 (1995) 262–264]. The protein appears to undergo an autocatalytic reaction to create the fluorophore [Heim et al., Proc. Natl. Acad. Sci. USA 91 (1994) 12501–12504] in a process involving cyclization of a Tyr⁶⁶ aa residue. Recently [Delagrave et al., Bio/Technology 13 (1995) 151–154], a combinatorial mutagenic strategy was targeted at aa 64 through 69, which spans the chromophore of A. victoria GFP, yielding a number of different mutants with redshifted fluorescence excitation spectra. One of these, RSGFP4, retains the characteristic green emission spectra (λ_max = 505 nm), but has a single excitation peak (λ_max = 490 nm). The fluorescence properties of RSGFP4 are similar to those of another naturally occurring GFP from the sea pansy, Renilla reniformis [Ward and Cormier, Photobiochem. Photobiol. 27 (1978) 389–396]. In the present study, we demonstrate by fluorescence microscopy that selective excitation of A. victoria GFP and RSGFP4 allows for spectral separation of each fluorescent signal, and provides the means to image these signals independently in a mixed population of bacteria or mammalian cells.     

Folding of green fluorescent protein and the cycle3 mutant

Although the correct folding of green fluorescent protein (GFP) is required for formation of the chromophore, it is known that wild-type GFP cannot mature efficiently in vivo in Escherichia coli at 37 degrees C or higher temperatures that the jellyfish in the Pacific Northwest have never experienced. Recently, by random mutagenesis by the polymerase chain reaction (PCR) method, a mutant called Cycle3 was constructed. This mutant had three mutations, F99S, M153T, and V163A, on or near the surface of the GFP molecule and was able to mature correctly even at 37 degrees C [Crameri et al. (1996) Nat. Biotechnol. 143, 315-319]. In the present study, we investigated the differences in their folding behavior in vitro. We observed the folding and unfolding reactions of both wild-type GFP and the Cycle3 mutant by using green fluorescence as an indicator of the formation of the native structure and examining hydrogen-exchange reactions by Fourier transform infrared spectroscopy. Both proteins showed unusually slow refolding and unfolding rates, and their refolding rates were almost identical under the native state at 25 and at 35 degrees C. On the other hand, aggregation studies in vitro showed that wild-type GFP had a strong tendency to aggregate, while the Cycle3 mutant did not. These results indicated that the ability to mature efficiently in vivo at 37 degrees C was not due to the improved folding and that reduced hydrophobicity on the surface of the Cycle3 mutant was a more critical factor for efficient maturation in vivo.     

GFP accumulation controlled by an auxin-responsive promoter as a non-destructive assay to monitor early auxin response

We have constructed transgenic Arabidopsis lines that contain a gene for green fluorescent protein (GFP) under the control of auxin-responsive domains A and B of the promoter from the pea PS-IAA4/5 gene. The chimeric transgene was named BA-mgfp5-ER. GFP was detected after the application of indole-3-acetic acid at concentrations as low as 100 nM in epidermal cells in the root elongation zone. The induction of the reporter gene was highly specific to auxin and was correlated with the auxin-induced change in epidermal cell shape. No GFP accumulation was observed in the lateral root meristem that was formed as a result of exogenous auxin application. These results suggest that auxin signals were transmitted through several distinct pathways depending on the cell type. The intensity of the GFP signal was strong enough to be observed through the plastic lid of the culture dish using a dissecting microscope, thereby enabling GFP expression to be monitored in an aseptic environment. Thus, the BA-mgfp5-ER transgenic plant can be a powerful tool for screening mutants that are defective in auxin signaling and the expression of early auxin-response genes.     

Comparison of morpholino based translational inhibition during the development of Xenopus laevis and Xenopus tropicalis

Morpholino (MO) based inhibition of translational initiation represents an attractive methodology to eliminate gene function during Xenopus development (Heasman et al., 2000). However, the degree to which a given target protein can be eliminated and the longevity of this effect during embryogenesis has not been documented. To examine the efficacy of MOs, we have used transgenic Xenopus lines that harbour known numbers of integrations of a GFP reporter under the control of the ubiquitous and highly expressed CMV promoter (Fig. 1a). In addition we have investigated the longevity of the inhibitory effect by using transgenic lines expressing GFP specifically in the lens of tadpoles. These transgenic lines represent the ideal control for the technique as the promoters are highly expressed and GFP can be easily detected by fluorescence and immunoblotting. Moreover, as GFP has no function in development, the levels of inhibition can be tested in an otherwise normal individual. Here we report that MOs are able to efficiently and specifically inhibit the translation of GFP in transgenic lines from Xenopus laevis and Xenopus tropicalis and the inhibitory effect is long-lived, lasting into the tadpole stages. genesis 30:110--113, 2001.     

Recent Advances in GFP Folding Reporter and Split-GFP Solubility Reporter Technologies. Application to Improving the Folding and Solubility of Recalcitrant Proteins from Mycobacterium tuberculosis

We have improved our green fluorescent protein (GFP) folding reporter technology [Waldo et al., (1999) Nat. Biotechnol. 17, 691–695] to evolve recalcitrant proteins from Mycobacterium tuberculosis. The target protein is inserted into the scaffolding of the GFP, eliminating false-positive artifacts caused by expression of truncated protein variants from internal cryptic ribosome binding sites in the target RNA. In parallel, we have developed a new quantitative fluorescent protein tagging and detection system based on micro-domains of GFP. This split-GFP system, which works both in vivo and in vitro, is amenable to high-throughput assays of protein expression and solubility [Cabantous et al., (2005) Nat. Biotechnol. 23, 102–107]. Together, the GFP folding reporter and split-GFP technologies offer a comprehensive system for manipulating and improving protein folding and solubility.     

A cellular 57 kDa protein binds to two regions of the internal translation initiation site of foot-and-mouth disease virus

A ribosome-associated 57 kDa protein from rabbit reticulocytes was linked to the internal translation initiation site of foot-and-mouth disease virus by mild UV-irradiation. Binding studies with different RNA fragments revealed that this protein interacts with two distinct sites within the translational control region. One site is located approximately 400 nucleotides upstream from the translational start codon and the second binding site could be confined to 60 nucleotides preceding this codon. Both sequences coincide with hairpin structures at the two opposite ends of a secondary structure model of the internal ribosomal entry site proposed by Pilipenko et al. [(1989) Nucleic Acids Res. 17, 5701-5711].     

Modeling of growth and energy metabolism of Pichia pastoris producing a fusion protein

A fusion protein composed of a cellulose binding domain from Neocallimastix patriciarum cellulase A and Candida antarctica lipase B (CBD-lipase) was produced by Pichia pastoris methanol utilization plus phenotype in high cell-density cultures. The genes expressing CBD-lipase were fused to the alpha-factor secretion signal sequence of Saccharomyces cerevisiae and placed under the control of the alcohol oxidase gene (AOX1) promoter. To control the repression and induction of AOX1 and oxygen demand at high cell density, a four-stage process was used. Batch growth on glycerol was used in the first step to provide biomass (28 g L^-1) while product formation was prevented due to repression of the AOX1. The second stage was exponential fed-batch growth on glycerol, which caused a slight increase of the enzyme alcohol oxidase activity due to derepression of the AOX1. This procedure resulted in smooth transition to exponential fed-batch growth on methanol, the third stage, in which the AOX1 was strongly induced. The fourth stage was constant fed-batch growth on methanol used to control the oxygen demand at the high cell density. A kinetic model was developed that could predict biomass growth and oxygen consumption in processes with and without oxygen-enriched air. With oxygen enrichment to 34% O₂ in the inlet air the methanol feed rate could be increased by 50% and this resulted in 14% higher final cell density (from 140 to 160 g L^-1 cell dry weight). The increased methanol feed rate resulted in a proportionally increased specific rate of product secretion to the medium. After an initial decrease, the synthesis capacity of the cell was kept constant throughout the cultivation, which made the product concentration increase almost constantly during the process. The kinetic model also describes how the low maintenance demand of P. pastoris compared with E. coli enables this organism to grow to such high cell densities.     

Current state and perspectives in modeling and control of human pluripotent stem cell expansion processes in stirred‐tank bioreactors

Implementation of model‐based practices for process development, control, automation, standardization, and validation are important factors for therapeutic and industrial applications of human pluripotent stem cells. As robust cultivation strategies for pluripotent stem cell expansion and differentiation have yet to be determined, process development could be enhanced by application of mathematical models and advanced control systems to optimize growth conditions. Therefore, it is important to understand both the potential of possible applications and the apparent limitations of existing mathematical models to improve pluripotent stem cell cultivation technologies. In the present review, the authors focus on these issues as they apply to stem cell expansion processes. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:355–364, 2017     

Death wins against life in a spatially extended model of the caspase-3/8 feedback loop

Apoptosis is an important physiological process which enables organisms to remove unwanted or damaged cells. A mathematical model of the extrinsic pro-apoptotic signaling pathway has been introduced by Eissing et al. (2007) and a bistable behavior with a stable death state and a stable life state of the reaction system has been established. In this paper, we consider a spatial extension of the extrinsic pro-apoptotic signaling pathway incorporating diffusion terms and make a model-based, numerical analysis of the apoptotic switch in the spatial dimension. For the parameter regimes under consideration it turns out that for this model diffusion homogenizes rapidly the concentrations which afterward are governed by the original reaction system. The activation of effector-caspase 3 depends on the space averaged initial concentration of pro-caspase 8 and pro-caspase 3 at the beginning of the process.     

X-ray structure of Cerulean GFP: a tryptophan-based chromophore useful for fluorescence lifetime imaging

The crystal structure of the cyan-fluorescent Cerulean green fluorescent protein (GFP), a variant of enhanced cyan fluorescent protein (ECFP), has been determined to 2.0 A. Cerulean bears an internal fluorophore composed of an indole moiety derived from Y66W, conjugated to the GFP-like imidazolinone ring via a methylene bridge. Cerulean undergoes highly efficient fluorescence resonance energy transfer (FRET) to yellow acceptor molecules and exhibits significantly reduced excited-state heterogeneity. This feature was rationally engineered in ECFP by substituting His148 with an aspartic acid [Rizzo et al. (2004) Nat. Biotechnol. 22, 445], rendering Cerulean useful for fluorescence lifetime imaging microscopy (FLIM). The X-ray structure is consistent with a single conformation of the chromophore and surrounding residues and may therefore provide a structural rationale for the previously described monoexponential fluorescence decay. Unexpectedly, the carboxyl group of H148D is found in a buried position, directly contacting the indole nitrogen of the chromophore via a bifurcated hydrogen bond. Compared to the similarly constructed ECFP chromophore, the indole group of Cerulean is rotated around the methylene bridge to adopt a cis-coplanar conformation with respect to the imidazolinone ring, resulting in a close edge-to-edge contact of the two ring systems. The double-humped absorbance spectrum persists in single-crystal absorbance measurements, casting doubt on the idea that ground state conformational heterogeneity forms the basis of the two overlapping transitions. At low pH, a blue shift in absorbance of 10-15 nm suggests a pH-induced structural transition that proceeds with a time constant of 47 (+/-2) min and is reversible. Possible interpretations in terms of chromophore isomerization are presented.     

Analysis of chemotactic bacterial distributions in population migration assays using a mathematical model applicable to steep or shallow attractant gradients

The mathematical model developed by Riveroet al. (1989,Chem. Engng Sci. 44, 2881–2897) is applied to literature data measuring chemotactic bacterial population distributions in response to steep as well as shallow attractant gradients. This model is based on a fundamental picture of the sensing and response mechanisms of individual bacterial cells, and thus relates individual cell properties such as swimming speed and tumbling frequency to population parameters such as the random motility coefficient and the chemotactic sensitivity coefficient. Numerical solution of the model equations generates predicted bacterial density and attractant concentration profiles for any given experimental assay. We have previously validated the mathematical model from experimental work involving a step-change in the attractant gradient (Fordet al., 1991Biotechnol. Bioengng.37, 647–660; For and Lauffenburger, 1991,Biotechnol. Bioengng,37, 661–672). Within the context of this experimental assay, effects of attractant diffusion and consumption, random motility, and chemotactic sensitivity on the shape of the profiles are explored to enhance our understanding of this complex phenomenon. We have applied this model to various other types of gradients with successful intepretation of data reported by Dalquistet al. (1972,Nature New Biol. 236, 120–123) forSalmonella typhimurum validating the mathematical model and supportin the involvement of high and low affinity receptors for serine chemotaxis by these cells.     

Phenol utilization by filamentous yeast Trichosporon cutaneum

The investigated strain Trichosporon cutaneum shows well expressed capability for metabolizing high concentrations of phenol, up to 1 g/l, utilizing it as the sole carbon source for the growth and development of the population. The data reported, prove the good perspectives for its application in protecting the environment from phenol pollution. No data about modelling the process of cultivation of Trichosporon cutaneum in phenol media is available in scientific literature up to now. The mathematical model, reported here, consists of two nonlinear differential equations, describing cell growth and substrate consumption. The unknown parameters are estimated following the method of Hooke and Jeeves. A number of simulation investigations are carried out. They prove the adequacy of the model and its applicability in further studies on the processes of growth and phenol uptake of Trichosporon cutaneum.     

Regulated nuclear polyadenylation of Xenopus albumin pre-mRNA.

Cytoplasmic regulation of the length of poly(A) on mRNA is a well-characterized process involved in translational control during development. In contrast, there is no direct in vivo evidence for regulation of the length of poly(A) added during nuclear pre-mRNA processing in somatic cells. We previously reported that Xenopus serum albumin [Schoenberg et al. (1989) Mol. Endocrinol. 3, 805-815] and transferrin [Pastori et al. (1992) J. Steroid Biochem. Mol. Biol. 42, 649-657], mRNA have exceptionally short poly(A) tails ranging from 12 to 17 residues, whereas vitellogenin mRNA has long poly(A). An RT-PCR protocol was adapted to determine the length of poly(A) added onto pre-mRNA, defined here as that species bearing the terminal intron. Using this assay we show that vitellogenin pre-mRNA has the same long poly(A) tail as mature vitellogenin mRNA. In contrast, albumin pre-mRNA has the same short poly(A) as found on fully-processed albumin mRNA. These results indicate that the short poly(A) tail on albumin mRNA results from regulation of poly(A) addition during nuclear 3' processing.