Validation of a high-throughput fermentation system based on online monitoring of biomass and fluorescence in continuously shaken microtiter plates

Background  

An advanced version of a recently reported high-throughput fermentation system with online measurement, called BioLector, and its validation is presented. The technology combines high-throughput screening and high-information content by applying online monitoring of scattered light and fluorescence intensities in continuously shaken microtiter plates. Various examples in calibration of the optical measurements, clone and media screening and promoter characterization are given.     

Engineering characterisation of a single well from 24-well and 96-well microtitre plates

The detailed engineering characterisation of shaken microtitre-plate bioreactors will enhance our understanding of microbial and mammalian cell culture in these geometries and will provide guidance on the scale-up of microwell results to laboratory and pilot scale stirred bioreactors. In this work computational fluid dynamics (CFD) is employed to provide a detailed characterisation of fluid mixing, energy dissipation rate and mass transfer in single well bioreactors from deep square 24-well and 96-well microtitre plates. The numerical predictions are generally found to be in good agreement with experimental observation of the fluid motion and measured values of the key engineering parameters. The CFD simulations have shown that liquid mixing is more intensive in 96-well than in 24-well bioreactors due to a significant axial component to the fluid velocity. Liquid motion is strongly dependent on the orbital shaking amplitude which generally has a greater impact than the shaking frequency. Average power consumptions of 70–100 W m⁻³ and 500–1000 W m⁻³, and overall mass transfer coefficient, k_La, values of 0.005–0.028 s⁻¹ and 0.056–0.10 s⁻¹ were obtained for 24-well and 96-well bioreactors respectively at an orbital shaking amplitude of 3 mm and shaking frequencies ranging from 500 rpm to 1500 rpm. The distribution of energy dissipation rates within each bioreactor showed these to be greatest at the walls of the well for both geometries. Batch culture kinetics of E. coli DH5 showed similar maximum specific growth rates and final biomass yields in shaken 24-well and shake flask bioreactors and in stirred miniature and 20 L bioreactors at matched k_La values. The CFD simulations thus give new insights into the local and overall engineering properties of microwell bioreactor geometries and further support their use as high throughput tools for the study and optimisation of microbial and mammalian cell culture kinetics at this scale.     

Oxygen transfer phenomena in 48-well microtiter plates: determination by optical monitoring of sulfite oxidation and verification by real-time measurement during microbial growth

Oxygen limitation is one of the most frequent problems associated with the application of shaking bioreactors. The gas-liquid oxygen transfer properties of shaken 48-well microtiter plates (MTPs) were analyzed at different filling volumes, shaking diameters, and shaking frequencies. On the one hand, an optical method based on sulfite oxidation was used as a chemical model system to determine the maximum oxygen transfer capacity (OTR(max)). On the other hand, the Respiration Activity Monitoring System (RAMOS) was applied for online measurement of the oxygen transfer rate (OTR) during growth of the methylotropic yeast Hansenula polymorpha. A proportionality constant between the OTR(max) of the biological system and the OTR(max) of the chemical system were indicated from these data, offering the possibility to transform the whole set of chemical data to biologically relevant conditions. The results exposed "out of phase" shaking conditions at a shaking diameter of 1 mm, which were confirmed by theoretical consideration with the phase number (Ph). At larger shaking diameters (2-50 mm) the oxygen transfer rate in MTPs shaken at high frequencies reached values of up to 0.28 mol/L/h, corresponding to a volumetric mass transfer coefficient (k(L)a) of 1,600 1/h. The specific mass transfer area (a) increases exponentially with the shaking frequency up to values of 2,400 1/m. On the contrary, the mass transfer coefficient (k(L)) is constant at a level of about 0.15 m/h over a wide range of shaking frequencies and shaking diameters. However, at high shaking frequencies, when the complete liquid volume forms a thin film on the cylindric wall of the well, the mass transfer coefficient (k(L)) increases linearly to values of up to 0.76 m/h. Essentially, the present investigation demonstrates that the 48-well plate outperforms the 96-well MTP and shake flasks at widely used operating conditions with respect to oxygen supply. The 48-well plates emerge, therefore, as an excellent alternative for microbial cultivation and expression studies combining the advantages of both the high-throughput 96-well MTP and the classical shaken Erlenmeyer flask.     

Multiplex bacterial growth monitoring in 24-well microplates using a dual optical sensor for dissolved oxygen and pH

Non-invasive, simultaneous optical monitoring of oxygen and pH during bacterial cultivation in 24-well microplates is presented using an integrated dual sensor for dissolved oxygen and pH values. The dual sensor is based on oxygen-sensitive organosilica microparticles and pH-sensitive microbeads from a polymethacrylate derivative embedded into a polyurethane hydrogel. The readout is based on a phase-domain fluorescence lifetime-based method referred to as modified frequency domain dual lifetime referencing using a commercially available detector system for 24-well microplates. The sensor was used for monitoring the growth of Pseudomonas putida bacterial cultures. The method is suitable for parallelized, miniaturized bioprocessing, and cell-based high-throughput screening applications.     

Macroscale versus microscale methods for physiological analysis of biofilms formed in 96-well microtiter plates

Microtiter plates with 96 wells have become one of the preferred platforms for biofilm studies mainly because they enable high-throughput assays. In this work, macroscale and microscale methods were used to study the impact of hydrodynamic conditions on the physiology and location of Escherichia coli JM109(DE3) biofilms formed in microtiter plates. Biofilms were formed in shaking and static conditions, and two macroscale parameters were assayed: the total amount of biofilm was measured by the crystal violet assay and the metabolic activity was determined by the resazurin assay. From the macroscale point of view, there were no statistically significant differences between the biofilms formed in static and shaking conditions. However, at a microscale level, the differences between both conditions were revealed using scanning electron microscopy (SEM). It was observed that biofilm morphology and spatial distribution along the wall were different in these conditions. Simulation of the hydrodynamic conditions inside the wells at a microscale was performed by computational fluid dynamics (CFD). These simulations showed that the shear strain rate was unevenly distributed on the walls during shaking conditions and that regions of higher shear strain rate were obtained closer to the air/liquid interface. Additionally, it was shown that wall regions subjected to higher shear strain rates were associated with the formation of biofilms containing cells of smaller size. Conversely, regions with lower shear strain rate were prone to have a more uniform spatial distribution of adhered cells of larger size. The results presented on this work highlight the wealth of information that may be gathered by complementing macroscale approaches with a microscale analysis of the experiments.     

Integrated optical sensing of dissolved oxygen in microtiter plates: a novel tool for microbial cultivation

Microtiter plates with integrated optical sensing of dissolved oxygen were developed by immobilization of two fluorophores at the bottom of 96-well polystyrene microtiter plates. The oxygen-sensitive fluorophore responded to dissolved oxygen concentration, whereas the oxygen-insensitive one served as an internal reference. The sensor measured dissolved oxygen accurately in optically well-defined media. Oxygen transfer coefficients, k(L)a, were determined by a dynamic method in a commercial microtiter plate reader with an integrated shaker. For this purpose, the dissolved oxygen was initially depleted by the addition of sodium dithionite and, by oxygen transfer from air, it increased again after complete oxidation of dithionite. k(L)a values in one commercial reader were about 10 to 40 h(-1). k(L)a values were inversely proportional to the filling volume and increased with increasing shaking intensity. Dissolved oxygen was monitored during cultivation of Corynebacterium glutamicum in another reader that allowed much higher shaking intensity. Growth rates determined from optical density measurement were identical to those observed in shaking flasks and in a stirred fermentor. Oxygen uptake rates measured in the stirred fermentor and dissolved oxygen concentrations measured during cultivation in the microtiter plate were used to estimate k(L)a values in a 96-well microtiter plate. The resulting values were about 130 h(-1), which is in the lower range of typical stirred fermentors. The resulting maximum oxygen transfer rate was 26 mM h(-1). Simulations showed that the errors caused by the intermittent measurement method were insignificant under the prevailing conditions.     

The baffled microtiter plate: Increased oxygen transfer and improved online monitoring in small scale fermentations

Most experiments in screening and process development are performed in shaken bioreactors. Today, microtiter plates are the preferred vessels for small‐scale microbial cultivations in high throughput, even though they have never been optimized for this purpose. To interpret the experimental results correctly and to obtain a base for a meaningful scale‐up, sufficient oxygen supply to the culture liquid is crucial. For shaken bioreactors this problem can generally be addressed by the introduction of baffles. Therefore, the focus of this study is to investigate how baffling and the well geometry affect the maximum oxygen transfer capacity (OTR_max) in microtiter plates. On a 48‐well plate scale, 30 different cross‐section geometries of a well were studied. It could be shown that the introduction of baffles into the common circular cylinder of a microtiter plate well doubles the maximum oxygen transfer capacity, resulting in values above 100 mmol/L/h (k_La > 600 1/h). To also guarantee a high volume for microbial cultivation, it is important to maximize the filling volume, applicable during orbital shaking. Additionally, the liquid height at the well bottom was examined, which is a decisive parameter for online‐monitoring systems such as the BioLector. This technology performs fiber‐optical measurements through the well bottom, therefore requires a constant liquid height at all shaking frequencies. Ultimately, a six‐petal flower‐shaped well geometry was shown to be the optimal solution taking into account all aforementioned criteria. With its favorable culture conditions and the possibility for unrestricted online monitoring, this novel microtiter plate is an efficient tool to gain meaningful results for interpreting and scaling‐up experiments in clone screening and bioprocess development. Biotechnol. Bioeng. 2009;103: 1118–1128. © 2009 Wiley Periodicals, Inc.     

Pyruvate flux distribution in NADH-oxidase-overproducing Lactococcus lactis strain as a function of culture conditions

The influence of growth conditions on product formation from glucose by Lactococcus lactis strain NZ9800 engineered for NADH-oxidase overproduction was examined. In aerobic batch cultures, a large production of acetoin and diacetyl was found at acidic pH under pH-unregulated conditions. However, pyruvate flux was mainly driven towards lactate production when these cells were grown under strictly pH-controlled conditions. A decreased NADH-oxidase overproduction accompanied the homolactic fermentation, suggesting that the cellular energy was used with preference to maintain cellular homeostasis rather than for NADH-oxidase overproduction. The end product formation and NADH-oxidase activity were also studied in cells grown in aerobic continuous cultures under acidic conditions. A homoacetic type of fermentation as well as a low NADH-oxidase overproduction were observed at low dilution rates. NADH-oxidase was efficiently overproduced as the dilution rate was increased and consequently metabolic flux through lactate dehydrogenase drastically decreased. Under these conditions the flux limitation via pyruvate dehydrogenase was relieved and this enzymatic complex accommodated most of the pyruvate flux. Pyruvate was also significantly converted to acetoin and diacetyl via alpha-acetolactate synthase. At higher dilution rates, acetate production declined and the cultures turned to mixed-acid fermentation. These results suggest that the need to maintain the cellular homeostasis influenced NADH-oxidase overproduction and consequently the end product formation from glucose in these engineered strains.     

Advancing 2D fluorescence online monitoring in microtiter plates by separating scattered light and fluorescence measurement,using a tunable emission monochromator

Online fluorescence monitoring has become a key technology in modern bioprocess development, as it provides in-depth process knowledge at comparably low costs. In particular, the technology is widely established for high-throughput microbioreactor cultivation systems, due to its noninvasive character. For microtiter plates, previously also multi-wavelength 2D fluorescence monitoring was developed. To overcome an observed limitation of fluorescence sensitivity, this study presents a modified spectroscopic setup, including a tunable emission monochromator. The new optical component enables the separation of the scattered and fluorescent light measurements, which allows for the adjustment of integration times of the charge-coupled device detector. The resulting increased fluorescence sensitivity positively affected the performance of principal component analysis for spectral data of Escherichia coli batch cultivation experiments with varying sorbitol concentration supplementation. In direct comparison with spectral data recorded at short integration times, more biologically consistent signal dynamics were calculated. Furthermore, during partial least square regression for E. coli cultivation experiments with varying glucose concentrations, improved modeling performance was observed. Especially, for the growth-uncoupled acetate concentration, a considerable improvement of the root-mean-square error from 0.25 to 0.17 g/L was achieved. In conclusion, the modified setup represents another important step in advancing 2D fluorescence monitoring in microtiter plates.     

Glucose-containing polymer rings enable fed-batch operation in microtiter plates with parallel online measurement of scattered light,fluorescence, dissolved oxygen tension,and pH

Bioprocesses operated in batch mode can induce adverse effects like overflow metabolism, substrate inhibition, osmotic inhibition, oxygen limitation, and catabolite repression. To avoid these adverse effects, fed-batch is the predominant operation mode in industrial production. Nevertheless, screening for optimal production strains is usually performed in microtiter plates and shake flasks operated in batch mode without any online monitoring. Recently, a polymer-based controlled-release fed-batch microtiter plate with stable glucose release characteristics was described. In this study, a glucose-containing polymer matrix was used to manufacture polymer rings that were placed at the bottom of a 48-well microtiter plate. Thereby, the liquid content of the well became accessible for optical measurement by the BioLector device. Reflections caused by the polymer ring were minimized by adjusting the scattered-light measurement position. Influences on the measurement of the dissolved oxygen tension and pH could be avoided by choosing appropriate polymer-ring geometries. These adjustments enabled parallel online measurement of scattered light, fluorescence, dissolved oxygen tension, and pH of Escherichia coli BL21 (DE3) fed-batch cultivations. The online monitoring and fed-batch operation capabilities of the fed-batch microtiter plate presented in this study finds optimal application in screenings and initial process development.     

Antibody production in Bacillus megaterium: strategies and physiological implications of scaling from microtiter plates to industrial bioreactors

Bacillus megaterium was used as an alternative high potential microbial production system for the production of antibody fragment D1.3 scFv. The aim of the study was to follow a holistic optimization approach from medium screening in small scale microtiter platforms, gaining deeper process understanding in the bioreactor scale and implementing advanced process strategies at larger scales (5-100 L). Screening and optimization procedures were supported by statistical design of experiments and a genetic algorithm approach. The process control relied on a soft-sensor for biomass estimation to establish a μ-oscillating time-dependent fed-batch strategy. Several cycles of growth phases and production phases, equal to starving phases, were performed in one production. Flow cytometry was used to monitor and characterize the dynamics of secretion and cell viability. Besides the biosynthesis of the product, secretion was optimized by an appropriate medium design considering different carbon sources, metal ions, (NH(4))(2)SO(4), and inductor concentrations. For bioprocess design, an adapted oscillating fed-batch strategy was conceived and successfully implemented at an industrially relevant scale of 100 L. In comparison to common methods for controlling fed-batch profiles, the developed process delivered increased overall productivities. Thereby measured process parameters such as growth stagnation or productivity fluctuations were directly linked to single cell or population behavior leading to a more detailed process understanding. Above all, the importance of single cell analysis as key scale-free tool to characterize and optimize recombinant protein production is highlighted, since this can be applied to all development stages independently of the cultivation platform.     

A theoretical and empirical investigation of delayed growth response in the continuous culture of bacteria

When the growth of bacteria in a chemostat is controlled by limiting the supply of a single essential nutrient, the growth rate is affected both by the concentration of this nutrient in the culture medium and by the amount of time that it takes for the chemical and physiological processes that result in the production of new biomass. Thus, although the uptake of nutrient by cells is an essentially instantaneous process, the addition of new biomass is delayed by the amount of time that it takes to metabolize the nutrient. Mathematical models that incorporate this "delayed growth response" (DGR) phenomenon have been developed and analysed. However, because they are formulated in terms of parameters that are difficult to measure directly, these models are of limited value to experimentalists. In this paper, we introduce a DGR model that is formulated in terms of measurable parameters. In addition, we provide for this model a complete set of criteria for determining persistence versus extinction of the bacterial culture in the chemostat. Specifically, we show that DGR plays a role in determining persistence versus extinction only under certain ranges of chemostat operating parameters. It is also shown, however, that DGR plays a role in determining the steady-state nutrient and bacteria concentrations in all instances of persistence. The steady state and transient behavior of solutions of our model is found to be in agreement with data that we obtained in growing Escherichia coli 23716 in a chemostat with glucose as a limiting nutrient. One of the theoretical predictions of our model that does not occur in other DGR models is that under certain conditions a large delay in growth response might actually have a positive effect on the bacteria's ability to persist.     

Online monitoring of oxygen in spinner flasks

We show the application of a novel optical on-line sensor fixed in spinner flasks for the online monitoring of dissolved O₂ concentrations during mammalian cell growth. Using this sensor that requires only minute changes to the flask to be made, we could determine the volumetric O₂ transfer coefficient as well as O₂ consumption rates. Under normal growth conditions the cells did not undergo O₂ limitation. Also, the transfer of O₂ from the atmosphere to the spinner flasks is influenced by the use of screw caps. The on-line measurement was further applied to determine the O₂ uptake rates which can then be used to monitor the metabolic state of the cells and also for online process monitoring.     

Diversity of microbial communities in open mixed culture fermentations: impact of the pH and carbon source

Anaerobic fermentation by an open mixed culture was investigated at different pH values (4–8.5) and with three substrates (glucose, glycerol and xylose). The populations established in each condition were assessed by denaturing gradient gel electrophoresis analysis of the 16S ribosomal RNA gene fragments. The fermentation pattern and the composition of the microbial population were also evaluated when operational variations were imposed (increase of substrate concentration or introduction of a second substrate). The experimental results demonstrated that at low and high pH values, a clearly different fermentation pattern was associated with the dominance of a specialised group of clostridiae. At intermediate pH values, the product spectrum was rather variable and seemed to be sensitive to variations in the microbial community. Different substrates resulted in the establishment of different microbial communities. When fed with a mixture of two substrates, mixotrophic microorganisms (capable of degrading both substrates) were found to overgrow the originally dominant specialists. Overall, the experiments have shown that some operational variables have a clear impact on the fermentation pattern and on the population established. However, a uniform relationship between the process characteristics (associated to a metabolic response) and the microbial population present is not always possible. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Influence of nematode age and culture conditions on morphological and physiological parameters in the bacterial vesicle of Steinernema carpocapsae (Nematoda: Steinernematidae)

Steinernema spp. third-stage infective juveniles (IJs) play a key role in the symbiotic partnership between these entomopathogenic nematodes and Xenorhabdus bacteria. Recent studies suggest that Steinernema carpocapsae IJs contribute to the nutrition and growth of their symbionts in the colonization site (vesicle) [Martens, E.C. and Goodrich-Blair, H., 2005. The S. carpocapsae intestinal vesicle contains a sub-cellular structure with which Xenorhabdus nematophila associates during colonization initiation. Cellular Microbiol. 7, 1723-1735.]. However, the morphological and physiological interactions between Xenorhabdus symbionts and Steinernema IJs are not understood in depth. This study was undertaken to assess the influence of culture conditions and IJ age on the structure, nutrition, and symbiont load (colonization level) of S. carpocapsae vesicles. Our observations indicate the vesicles of axenic IJs are shorter and wider than those of colonized IJs. Moreover, as colonized IJs age the vesicle becomes shorter and narrower and bacterial load declines. The colonization proficiency of several bacterial metabolic mutants was compared between two cultivation conditions: in vitro on lipid agar and in vivo in Galleria mellonella insects. Colonization defects were generally less severe in IJs cultivated in vivo versus those cultivated in vitro. However, IJs from both cultivation conditions exhibited similar declining bacterial load over time. These results suggest that although the vesicle forms in the absence of bacteria, the presence of symbionts within the vesicle may influence its fine structure. Moreover, these studies provide further evidence in support of the concept that the conditions under which steinernematid nematodes are cultivated and stored affect the nutritive content of the vesicle and the bacterial load, and therefore have an impact on the quality of the nematodes for their application as biological control agents.     

Selective labeling and monitoring pH changes of lysosomes in living cells with fluorogenic pH sensors

New BODIPY-based pH probes have been designed with excitation and emission wavelengths suitable for fluorescence microscopy and flow cytometry. These pH probes are cell-permeable, selectively label lysosomes, and can be used for noninvasive monitoring of lysosomal pH changes during physiological and pathological processes.     

Influence of temperature rise on kinetic parameters during batch propagation of Kluyveromyces fragilis in cheese whey under ambient conditions

Three 5 l working volume fermenters were used to investigate the growth of the yeast Kluyveromyces fragilis in acid cheese whey under ambient temperature in order to assess the specific growth rate and yield, the lactose and oxygen uptake rates during the various phases of batch culture, the effect of increasing temperature on the various kinetic parameters, and the need for a cooling unit for single cell production batch systems. The initial dissolved oxygen in the medium was 5.5 mg l^–1 and the pH was maintained at 4.5. The observed lag phase, specific growth rate and maximum cell number were 4 h, 0.2 h^–1 and 8.4 × 10⁸ cells ml^–1, respectively. About 99% of the lactose in cheese whey was utilized within 20 h, 85% during the exponential growth phase. The specific lactose utilization rates by K. fragilis were 0.20 × 10^–12, 1.457 × 10^–12, 0.286 × 10^–12 and 0.00 g lactose cell^–1 h^–1, for the lag, exponential, stationary and death phases, respectively. The dissolved oxygen concentration in the medium decreased as the cell number increased. The lowest oxygen concentration of 1.2 mg l^–1 was observed during the stationary phase. The volumetric oxygen transfer coefficient was 0.41 h^–1 and the specific oxygen uptake rates were 0.32 × 10^–12, 2.14 × 10^–12, 0.51 × 10^–12 and 0.003 × 10^–12 mg O₂ cell^–1 h^–1, for the lag, exponential, stationary and death phases, respectively. The maximum temperature recorded for the medium was 33 °C, indicating that a cooling unit for batch production of single cell protein at ambient temperature is not needed for this type of bioreactor. The increase in medium temperature affected the cell growth and the lactose and oxygen uptake rates.     

A novel approach for estimating growth phases and parameters of bacterial population in batch culture

Using mathematical analysis, a new method has been developed for studying the growth kinetics of bacterial populations in batch culture. First, sampling data were smoothed with the spline interpolation method. Second, the instantaneous rates were derived by numerical differential techniques and finally, the derived data were fitted with the Gaussian function to obtain growth parameters. We named this the Spline-Numerical-Gaussian or SNG method. This method yielded more accurate estimates of the growth rates of bacterial populations and new parameters. It was possible to divide the growth curve into four different but continuous phases based on changes in the instantaneous rates. The four phases are the accelerating growth phase, the constant growth phase, the decelerating growth phase and the declining phase. Total DNA content was measured by flow cytometry and varied depending on the growth phase. The SNG system provides a very powerful tool for describing the kinetics of bacterial population growth. The SNG method avoids the unrealistic assumptions generally used in the traditional growth equations.     

A novel approach for estimating growth phases and parameters of bacterial population in batch culture

Traditionally, microbiologists divided bacterial growth in batch cultures into lag, exponential, station-ary and death phases[1], following the Logistic equa-tion that has been applied to the growth of human populations. The growth curves can always be ch…