Identifying validity evidence for uncertainty tolerance scales: A systematic review.

PURPOSE: Uncertainty tolerance (UT) is increasingly valued as a medical graduate attribute and broadly measured among medical student populations. However, the validity evidence underpinning UT scale implementation has not been summarised across studies. The present work evaluates UT scale validity evidence to better inform when, why and how UT scales ought to be used and to identify remaining validity evidence gaps. METHODS: A literature search for psychometric studies of UT scales was completed in 2022. Records were included if they implemented one of the four most cited UT scales (i.e. Physicians' Reactions to Uncertainty scale 1990 [PRU1990] or 1995 [PRU1995], Tolerance for Ambiguity [TFA] scale or Tolerance of Ambiguity in Medical Students and Doctors scale [TAMSAD]) in a population of physicians and/or medial students and presented validity evidence according to the Standards for Educational and Psychological Testing framework. Included studies were rated and analysed according to evidence for test content, response processes, internal structure, relations to other variables and consequences of testing. RESULTS: Among the investigated scales, 'relations to other variables' and 'internal structure' were the most commonly reported forms of validity evidence. No evidence of 'response processes' or 'consequences of testing' was identified. Overall, the PRU1990 and PRU1995 demonstrated the strongest validity evidence, although evidence primarily related to physician populations. CONCLUSIONS: None of the studied scales demonstrated evidence for all five sources of validity. Future research would benefit from assessing validity evidence for 'response processes' and 'consequences of testing' among physicians and medical students at different training/career stages to better understand UT construct conceptualisation in these populations. Until further and stronger validity evidence for UT scales is established, we caution against implementing UT scales outside of research settings (e.g. for higher stakes decision making).

Reliability of Uncertainty Tolerance Scales Implemented Among Physicians and Medical Students: A Systematic Review and Meta-Analysis.

PURPOSE: Uncertainty tolerance (UT) is a construct describing individuals' perceptions of, and responses to, uncertainty across their cognition, emotion, and behavior. Various UT scales have been designed for physician and medical student populations. However, links between UT and other variables (e.g., training stages) are inconsistent, raising concerns about scale reliability and validity. As reliability is a precondition for validity, a necessary first step in assessing UT scales' efficacy is evaluating their reliability. Accordingly, the authors conducted a meta-analysis of the reliability of UT scales designed for, and implemented among, physician and medical student populations. METHOD: In 2020, the authors searched 4 electronic databases alongside a citation search of previously identified UT scales. They included English-language, peer-reviewed studies that implemented UT scales in physician and/or medical student populations and reported reliability evidence. A meta-analysis of studies' Cronbach's alphas evaluated aggregated internal consistency across studies; subgroup analyses evaluated UT scales by named scale, population, and item characteristics. RESULTS: Among 4,124 records screened, 35 studies met the inclusion criteria, reporting 75 Cronbach's alphas. Four UT scales appeared in at least 3 included studies: Physicians' Reactions to Uncertainty scale 1990 (PRU1990) and 1995 (PRU1995) versions, Tolerance for Ambiguity scale (TFA), and Tolerance of Ambiguity in Medical Students and Doctors scale (TAMSAD). The scores from these scales ranged in reliability from very good (PRU1990: 0.832, PRU1995: 0.818) to respectable (TFA: 0.761, TAMSAD: 0.711). Aggregated internal consistency was significantly higher ( P < .001) among physicians (0.797) than medical students (0.711). CONCLUSIONS: UT scales generally demonstrated respectable internal consistency when administered among physicians and medical students, yet the reliability among medical students was significantly lower. The authors caution against using UT scores for decision-making purposes (e.g., applicant selection, program evaluation), especially among medical student populations. Future research should explore the reasons underlying these observed population differences.

Kinetic modeling of countercurrent saccharification.

BACKGROUND: Countercurrent saccharification is a promising way to minimize enzyme loading while obtaining high conversions and product concentrations. However, in countercurrent saccharification experiments, 3-4 months are usually required to acquire a single steady-state data point. To save labor and time, simulation of this process is necessary to test various reaction conditions and determine the optimal operating point. Previously, a suitable kinetic model for countercurrent saccharification has never been reported. The Continuum Particle Distribution Modeling (CPDM) satisfactorily predicts countercurrent fermentation using mixed microbial cultures that digest various feedstocks. Here, CPDM is applied to countercurrent enzymatic saccharification of lignocellulose. RESULTS: CPDM was used to simulate multi-stage countercurrent saccharifications of a lignocellulose model compound (α-cellulose). The modified HCH-1 model, which accurately predicts long-term batch saccharification, was used as the governing equation in the CPDM model. When validated against experimental countercurrent saccharification data, it predicts experimental glucose concentrations and conversions with the average errors of 3.5% and 4.7%, respectively. CPDM predicts conversion and product concentration with varying enzyme-addition location, total stage number, enzyme loading, liquid residence time (LRT), and solids loading rate (SLR). In addition, countercurrent saccharification was compared to batch saccharification at the same conversion, product concentration, and reactor volume. Results show that countercurrent saccharification is particularly beneficial when the product concentration is low. CONCLUSIONS: The CPDM model was used to simulate multi-stage countercurrent saccharification of α-cellulose. The model predictions agreed well with the experimental glucose concentrations and conversions. CPDM prediction results showed that the enzyme-addition location, enzyme loading, LRT, and SLR significantly affected the glucose concentration and conversion. Compared to batch saccharification at the same conversion, product concentration, and reactor volume, countercurrent saccharification is particularly beneficial when the product concentration is low.

Development of modified HCH-1 kinetic model for long-term enzymatic cellulose hydrolysis and comparison with literature models.

BACKGROUND: Enzymatic hydrolysis is a major step for cellulosic ethanol production. A thorough understanding of enzymatic hydrolysis is necessary to help design optimal conditions and economical systems. The original HCH-1 (Holtzapple-Caram-Humphrey-1) model is a generalized mechanistic model for enzymatic cellulose hydrolysis, but was previously applied only to the initial rates. In this study, the original HCH-1 model was modified to describe integrated enzymatic cellulose hydrolysis. The relationships between parameters in the HCH-1 model and substrate conversion were investigated. Literature models for long-term (> 48 h) enzymatic hydrolysis were summarized and compared to the modified HCH-1 model. RESULTS: A modified HCH-1 model was developed for long-term (> 48 h) enzymatic cellulose hydrolysis. This modified HCH-1 model includes the following additional considerations: (1) relationships between coefficients and substrate conversion, and (2) enzyme stability. Parameter estimation was performed with 10-day experimental data using α-cellulose as substrate. The developed model satisfactorily describes integrated cellulose hydrolysis data taken with various reaction conditions (initial substrate concentration, initial product concentration, enzyme loading, time). Mechanistic (and semi-mechanistic) literature models for long-term enzymatic hydrolysis were compared with the modified HCH-1 model and evaluated by the corrected version of the Akaike information criterion. Comparison results show that the modified HCH-1 model provides the best fit for enzymatic cellulose hydrolysis. CONCLUSIONS: The HCH-1 model was modified to extend its application to integrated enzymatic hydrolysis; it performed well when predicting 10-day cellulose hydrolysis at various experimental conditions. Comparison with the literature models showed that the modified HCH-1 model provided the best fit.

Economic improvement of continuous pharmaceutical production via the optimal control of a multifeed bioreactor.

Projections on the profitability of the pharmaceutical industry predict a large amount of growth in the coming years. Stagnation over the last 20 years in product development has led to the search for new processing methods to improve profitability by reducing operating costs or improving process productivity. This work proposes a novel multifeed bioreactor system composed of independently controlled feeds for substrate(s) and media used that allows for the free manipulation of the bioreactor supply rate and substrate concentrations to maximize bioreactor productivity and substrate utilization while reducing operating costs. The optimal operation of the multiple feeds is determined a priori as the solution of a dynamic optimization problem using the kinetic models describing the time-variant bioreactor concentrations as constraints. This new bioreactor paradigm is exemplified through the intracellular production of beta-carotene using a three feed bioreactor consisting of separate glucose, ethanol and media feeds. The performance of a traditional bioreator with a single substrate feed is compared to that of a bioreactor with multiple feeds using glucose and/or ethanol as substrate options. Results show up to a 30% reduction in the productivity with the addition of multiple feeds, though all three systems show an improvement in productivity when compared to batch production. Additionally, the breakeven selling price of beta-carotene is shown to decrease by at least 30% for the multifeed bioreactor when compared to the single feed counterpart, demonstrating the ability of the multifeed reactor to reduce operating costs in bioreactor systems. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:902-912, 2017.

A novel method for furfural recovery via gas stripping assisted vapor permeation by a polydimethylsiloxane membrane.

Furfural is an important platform chemical with a wide range of applications. However, due to the low concentration of furfural in the hydrolysate, the conventional methods for furfural recovery are energy-intensive and environmentally unfriendly. Considering the disadvantages of pervaporation (PV) and distillation in furfural separation, a novel energy-efficient 'green technique', gas stripping assisted vapor permeation (GSVP), was introduced in this work. In this process, the polydimethylsiloxane (PDMS) membrane was prepared by employing water as solvent. Coking in pipe and membrane fouling was virtually non-existent in this new process. In addition, GSVP was found to achieve the highest pervaporation separation index of 216200 (permeate concentration of 71.1 wt% and furfural flux of 4.09 kg m(-2) h(-1)) so far, which was approximately 2.5 times higher than that found in pervaporation at 95°C for recovering 6.0 wt% furfural from water. Moreover, the evaporation energy required for GSVP decreased by 35% to 44% relative to that of PV process. Finally, GSVP also displayed more promising potential in industrial application than PV, especially when coupled with the hydrolysis process or fermentation in biorefinery industry.

Preparation of poly(phthalazinone-ether-sulfone) sponge-like ultrafiltration membrane.

Poly(phthalazinone-ether-sulfone) (PPES) polymer is a relatively newly developed material with a bis(4-fluorodiphenyl) sulfone group. The formation of the PPES membrane by wet-phase inversion can proceed according to a slow or fast gelation method. These formation mechanisms were studied experimentally. The resulting membrane morphology was investigated using both optical and scanning electron micrography. The effects of PPES concentration and two additives, polyvinylpyrrolidone (PVP) and oxalic acid (OA), on the apparent viscosity and gelation rate of PPESK/NMP solutions and membrane performance have also been investigated. It was found that the gelation rate is important to obtain a sponge-like membrane structure, however favored by a fast gelation rate. The membrane obtained by a fast gelation rate showed a high pure water flux and rejection of bovine serum albumin (BSA), contrary to previous findings. On the basis of the experimental results, the actual membrane structure and pure water flux were related, and in agreement with the optical micrograph and gelation rate, respectively. The current results provide a fundamental insight in this novel copolymer, useful in future applications, especially in the membrane formation process.

Potential of mean force for separation of the repeating units in cellulose and hemicellulose.

As a first step toward understanding the energetics of removal of cello-oligomers from the cellulose surface, we have performed umbrella sampling calculations to determine the free energy required for separation of repeating units of cellulose and hemicellulose from each other. Molecular dynamics (MD) simulations were performed for both the stacked and non-stacked arrangements of the cellobiose pair system and the xylobiose pair system. These stacked and non-stacked arrangements were taken as representative systems for the crystalline and amorphous domains in cellulose and hemicellulose. In addition, similar calculations were also carried out to determine the energetics involved in the separation of the cellobiose-xylobiose molecule pair in the non-stacked arrangement. The potential of mean force profiles exhibit a single minimum in all cases and are qualitatively similar. Our results show that the location of the minimum as well as the depth of the well can be correlated with the size of the disaccharide molecules.

Characterization and selectivity studies of molecular imprinted membranes of Puerarin using scanning electron microscopy.

Molecular-imprinted membranes of Puerarin were prepared by phase inversion technique with acrylonitrile-acrylic acid copolymer (P (AN-co-AA)). To characterize P (AN-co-AA), ubbelohde viscometer was used to measure its viscosity-molecular weight. P (AN-co-AA) with different molecular weight was used to prepare membranes. The copolymer-dimethyl sulfoxide solution with Puerarin (PU) template was coagulated in water at various temperatures. The increase in P (AN-co-AA) molecular weight and the decrease in coagulation temperature caused an increase in PU recognition property of the resultant membrane. The PU imprinted membrane prepared with P (AN-co-AA) showed good selective ability to PU. The purity of PU increased from 56.51 to 98.41 wt%. Surface and cross-section morphology of the membranes were analyzed by using scanning electron micrograph. High-performance liquid chromatography was used for the quantification of Puerarin in isolated fraction.

Model-based fed-batch for high-solids enzymatic cellulose hydrolysis.

While many kinetic models have been developed for the enzymatic hydrolysis of cellulose, few have been extensively applied for process design, optimization, or control. High-solids operation of the enzymatic hydrolysis of lignocellulose is motivated by both its operation decreasing capital costs and increasing product concentration and hence separation costs. This work utilizes both insights obtained from experimental work and kinetic modeling to develop an optimization strategy for cellulose saccharification at insoluble solids levels greater than 15% (w/w), where mixing in stirred tank reactors (STRs) becomes problematic. A previously developed model for batch enzymatic hydrolysis of cellulose was modified to consider the effects of feeding in the context of fed-batch operation. By solving the set of model differential equations, a feeding profile was developed to maintain the insoluble solids concentration at a constant or manageable level throughout the course of the reaction. Using this approach, a stream of relatively concentrated solids (and cellulase enzymes) can be used to increase the final sugar concentration within the reactor without requiring the high initial levels of insoluble solids that would be required if the operation were performed in batch mode. Experimental application in bench-scale STRs using a feed stream of dilute acid-pretreated corn stover solids and cellulase enzymes resulted in similar cellulose conversion profiles to those achieved in batch shake-flask reactors where temperature control issues are mitigated. Final cellulose conversions reached approximately 80% of theoretical for fed-batch STRs fed to reach a cumulative solids level of 25% (w/w) initial insoluble solids.

Soluble and insoluble solids contributions to high-solids enzymatic hydrolysis of lignocellulose.

The rates and extents of enzymatic cellulose hydrolysis of dilute acid pretreated corn stover (PCS) decline with increasing slurry concentration. However, mass transfer limitations are not apparent until insoluble solids concentrations approach 20% w/w, indicating that inhibition of enzyme hydrolysis at lower solids concentrations is primarily due to soluble components. Consequently, the inhibitory effects of pH-adjusted pretreatment liquor on the enzymatic hydrolysis of PCS were investigated. A response surface methodology (RSM) was applied to empirically model how hydrolysis performance varied as a function of enzyme loading (12-40 mg protein/g cellulose) and insoluble solids concentration (5-13%) in full-slurry hydrolyzates. Factorial design and analysis of variance (ANOVA) were also used to assess the contribution of the major classes of soluble components (acetic acid, phenolics, furans, sugars) to total inhibition. High sugar concentrations (130 g/L total initial background sugars) were shown to be the primary cause of performance inhibition, with acetic acid (15 g/L) only slightly inhibiting enzymatic hydrolysis and phenolic compounds (9 g/L total including vanillin, syringaldehyde, and 4-hydroxycinnamic acid) and furans (8 g/L total of furfural and hydroxymethylfurfural, HMF) with only a minor effect on reaction kinetics. It was also demonstrated that this enzyme inhibition in high-solids PCS slurries can be approximated using a synthetic hydrolyzate composed of pure sugars supplemented with a mixture of acetic acid, furans, and phenolic compounds, which indicates that generally all of the reaction rate-determining soluble compounds for this system can be approximated synthetically.

Tubular bioreactor models that include Onsager-Curie scalar cross-phenomena to describe stress-dependent rates of cell proliferation.

The theory of heterogeneous catalysis in chemical reactors is employed to simulate laminar flow through tubes at large mass transfer Peclet numbers in which anchorage-dependent cells (i) adhere to a protein coating on the inner surface at r=R(wall), (ii) receive nutrients and oxygen from an aqueous medium via transverse diffusion toward the active wall, and (iii) proliferate in the presence of viscous shear at the cell/aqueous-medium interface. This process is modeled as convective diffusion in cylindrical coordinates with chemical reaction at the boundary, where chemical reaction describes the rate of nutrient consumption. The formalism of irreversible thermodynamics is employed to describe an unusual coupling between viscous shear, or velocity gradients at the cell/aqueous-medium interface, and rates of nutrient consumption. Linear transport laws in chemically reactive systems that obey Curie's theorem predict the existence of cross-phenomena between fluxes (i.e., scalar reaction rates) and driving forces (i.e., 2nd-rank velocity gradient tensor) whose tensorial ranks differ by an even integer-in this case, two. This methodology for stress-dependent chemical reactions yields an additional zeroth-order contribution, via the magnitude of the velocity gradient tensor, to heterogeneous kinetic rate expressions because nutrient consumption and cell proliferation are stress-sensitive. Computer simulations of nutrient consumption suggest that bioreactor designs should consider stress-sensitive reactions when the shear-rate-based Damköhler number (i.e., defined for the first time in this study as the stress-dependent zeroth-order rate of nutrient consumption relative to the rate of nutrient diffusion toward active cells adhered to the tube wall) is greater than 10-20% of the stress-free Damköhler number. Models of bioreactor performance are presented for simple 1st-order, simple 2nd-order, and complex chemical kinetic rate expressions, where the latter considers adsorption/desorption equilibria via the Fowler-Guggenheim modification of the Langmuir isotherm for cell-protein docking on active sites, accompanied by cell-cell attraction. Stress sensitivity is magnified in physically realistic cell-based tubular bioreactors with complex stress-free kinetic rate expressions relative to simulations with simple 1st- and 2nd-order kinetics.

Cytotoxicity of aggregated fullerene C60 particles on CHO and MDCK cells.

The cytotoxicity of fullerene C60 particles on two mammalian cell lines, i.e. the Chinese hamster ovary (CHO) cells and the Madin-Darby canine kidney (MDCK) cells, has been investigated. Although innate fullerene particles have a very low solubility in deionized (DI) water, these particles can be dissolved in the tetrahydrofuran (THF) solvent at a great value. Further, the dissolved fullerene particles in the THF solvent could be extracted into a DI water solution at a significantly increased solubility. The formation of fullerene particle aggregates is believed to be the cause of the increased solubility. Results presented here show that once the concentration of the fullerene aggregates reaches a certain level, the cells start to die. The lethal dosage LD50, which is defined as the lowest fullerene concentration that results in a 50% cell death within 24 h, has been determined. Furthermore, the percentage of cell mortality increased with increasing fullerene concentration and incubation time yielding a negative effect on cell viability. These results, illustrated by atomic force microscopy (AFM), dynamic light scattering (DLS) and other microscopic techniques, will help to better understand the side effects of fullerene particles in mammalian cells.

Prediction of N-linked glycan branching patterns using artificial neural networks.

A model was developed for novel prediction of N-linked glycan branching pattern classification for CHO-derived N-linked glycoproteins. The model consists of 30 independent recurrent neural networks and uses predicted quantities of secondary structure elements and residue solvent accessibility as an input vector. The model was designed to predict the major component of a heterogeneous mixture of CHO-derived glycoforms of a recombinant protein under normal growth conditions. Resulting glycosylation prediction is classified as either complex-type or high mannose. The incorporation of predicted quantities in the input vector allowed for theoretical mutant N-linked glycan branching predictions without initial experimental analysis of protein structures. Primary amino acid sequence data were effectively eliminated from the input vector space based on neural network prediction analyses. This provided further evidence that localized protein secondary structure elements and conformational structure may play more important roles in determining glycan branching patterns than does the primary sequence of a polypeptide. A confidence interval parameter was incorporated into the model to enable identification of false predictions. The model was further tested using published experimental results for mutants of the tissue-type plasminogen activator protein [J. Wilhelm, S.G. Lee, N.K. Kalyan, S.M. Cheng, F. Wiener, W. Pierzchala, P.P. Hung, Alterations in the domain structure of tissue-type plasminogen activator change the nature of asparagine glycosylation. Biotechnology (N.Y.) 8 (1990) 321-325].

Modeling intrinsic kinetics of enzymatic cellulose hydrolysis.

A multistep approach was taken to investigate the intrinsic kinetics of the cellulase enzyme complex as observed with hydrolysis of noncrystalline cellulose (NCC). In the first stage, published initial rate mechanistic models were built and critically evaluated for their performance in predicting time-course kinetics, using the data obtained from enzymatic hydrolysis experiments performed on two substrates: NCC and alpha-cellulose. In the second stage, assessment of the effect of reaction intermediates and products on intrinsic kinetics of enzymatic hydrolysis was performed using NCC hydrolysis experiments, isolating external factors such as mass transfer effects, physical properties of substrate, etc. In the final stage, a comprehensive intrinsic kinetics mechanism was proposed. From batch experiments using NCC, the time-course data on cellulose, cello-oligosaccharides (COS), cellobiose, and glucose were taken and used to estimate the parameters in the kinetic model. The model predictions of NCC, COS, cellobiose, and glucose profiles show a good agreement with experimental data generated from hydrolysis of different initial compositions of substrate (NCC supplemented with COS, cellobiose, and glucose). Finally, sensitivity analysis was performed on each model parameter; this analysis provides some insights into the yield of glucose in the enzymatic hydrolysis. The proposed intrinsic kinetic model parametrized for dilute cellulose systems forms a basis for modeling the complex enzymatic kinetics of cellulose hydrolysis in the presence of limiting factors offered by substrate and enzyme characteristics.

Quantifying the metabolic capabilities of engineered Zymomonas mobilis using linear programming analysis.

BACKGROUND: The need for discovery of alternative, renewable, environmentally friendly energy sources and the development of cost-efficient, "clean" methods for their conversion into higher fuels becomes imperative. Ethanol, whose significance as fuel has dramatically increased in the last decade, can be produced from hexoses and pentoses through microbial fermentation. Importantly, plant biomass, if appropriately and effectively decomposed, is a potential inexpensive and highly renewable source of the hexose and pentose mixture. Recently, the engineered (to also catabolize pentoses) anaerobic bacterium Zymomonas mobilis has been widely discussed among the most promising microorganisms for the microbial production of ethanol fuel. However, Z. mobilis genome having been fully sequenced in 2005, there is still a small number of published studies of its in vivo physiology and limited use of the metabolic engineering experimental and computational toolboxes to understand its metabolic pathway interconnectivity and regulation towards the optimization of its hexose and pentose fermentation into ethanol. RESULTS: In this paper, we reconstructed the metabolic network of the engineered Z. mobilis to a level that it could be modelled using the metabolic engineering methodologies. We then used linear programming (LP) analysis and identified the Z. mobilis metabolic boundaries with respect to various biological objectives, these boundaries being determined only by Z. mobilis network's stoichiometric connectivity. This study revealed the essential for bacterial growth reactions and elucidated the association between the metabolic pathways, especially regarding main product and byproduct formation. More specifically, the study indicated that ethanol and biomass production depend directly on anaerobic respiration stoichiometry and activity. Thus, enhanced understanding and improved means for analyzing anaerobic respiration and redox potential in vivo are needed to yield further conclusions for potential genetic targets that may lead to optimized Z. mobilis strains. CONCLUSION: Applying LP to study the Z. mobilis physiology enabled the identification of the main factors influencing the accomplishment of certain biological objectives due to metabolic network connectivity only. This first-level metabolic analysis model forms the basis for the incorporation of more complex regulatory mechanisms and the formation of more realistic models for the accurate simulation of the in vivo Z. mobilis physiology.

Optimization of fed-batch parameters and harvest time of CHO cell cultures for a glycosylated product with multiple mechanisms of inactivation.

Optimization of fed-batch feeding parameters was explored for a system with multiple mechanisms of product inactivation. In particular, two separate mechanisms of inactivation were identified for the recombinant tissue-type activator (r-tPA) protein. Dynamic inactivation models were written to describe particular r-tPA glycoform inactivation in the presence and absence of free-glucose. A glucose-independent inactivation mechanism was identified, and inactivation rate constants were found dependent upon the presence of glycosylation of r-tPA at N184. Inactivation rate constants of the glucose-dependent mechanism were not affected by glycosylation at N184. Fed-batch optimization was performed for r-tPA production by CHO cell culture in a stirred-tank reactor with glucose, glutamine and asparagine feed. Feeding profiles in which culture supernatant concentrations of free-glucose and amino acids (combined glutamine and asparagine) were used as control variables, were evaluated for a wide variety of set points. Simulation results for a controlled feeding strategy yielded an optimum at set points of 1.51 g L(-1) glucose and 1.18 g L(-1) of amino acids. Optimization was also performed in absence of metabolite control using fixed feed-flow rates initiate during the exponential growth phase. Fixed feed-flow results displayed a family of optimum solutions along a mass flow rate ratio of 3.15 of glucose to amino acids. Comparison of the two feeding strategies showed a slight advantage of rapid feeding at a fixed flow rate as opposed to metabolite control for a product with multiple mechanisms of inactivation.

Development of a culture sub-population induction model: signaling pathways synergy and taxanes production by Taxus canadensis.

Cell cultures of Taxus canadensis were subjected to exogenously applied ethylene (ET) hormone and methyl jasmonate (MJ) elicitation in factorial design experiments. Levels of extracellular taxanes, including paclitaxel, were used with principal component analysis for fault detection and real-coded genetic algorithms for parameter optimization to construct a culture sub-population induction model. Culture sub-populations were identified by the model as (1) uninduced, (2) induced to unilateral function of the ET-signaling pathway, and (3) induced to cooperation between jasmonic acid (JA)- and ET-signaling pathways. Comprehensive model results suggested greater rates of cellular induction (resulting in exogenous taxane production) by ET gas as opposed to MJ elicitation. However, cellular induction of ET-signaling pathway genes increased the rate of induction of JA-signaling pathway genes by orders of magnitude. In addition, model results showed that induction of genes leading to extracellular production of the simple taxane 10-deacetylbaccatin III was regulated by the unilateral ET-signaling pathway. However, it was suggested that further processing of this simple taxane to complex taxane structures, such as paclitaxel, required further gene induction by the JA-signaling pathway. Thus, production rate constants of exogenous complex taxanes were predicted to be an order of magnitude lower than that for the simple taxane 10-deacetylbaccatin III. The fraction of the cell culture sub-population displaying unilateral ET-signaling pathway gene induction was found inversely proportional to levels of MJ elicitation. When coupled with simple non-growth product models, levels of all extracellular taxanes were effectively predicted using the culture sub-population induction model.

Variable site-occupancy classification of N-linked glycosylation using artificial neural networks.

A novel neural-network-based model has been developed for the prediction of N-linked glycosylation characteristics related to glycosylation site-occupancy. Intracellular oligosaccharide transfer to a polypeptide is known to be either robust or dependent upon culture conditions during pharmaceutical production. This glycan attachment is classified by the model as robust or variable and is based on an input of the polypeptide primary sequence around the site of glycosylation. The glycosylation model utilizes multiple recurrent neural networks followed by a perceptron classifier. The input length of the polypeptide chain around the site of glycosylation (glycosylation window) was optimized through multiple independent training sessions. Incorporation of five residues prior (n - 5) to the site of glycosylation (n) and four residues beyond (n + 4) the glycan attachment site led to optimal network performance. The size of the glycosylation window for site-occupancy determination is much larger than has been previously reported. This model was developed to evaluate the effects of theoretical polypeptide mutations on glycosylation site-occupancy characteristics. Following correct prediction of the model testing data set, 20 independent networks were used to predict site-occupancy characteristics of wild-type and mutants of the rabies virus glycoprotein (rgp). Simulation results strongly correlated with previously published experimental results (Kasturi, L.; Hegang, C.; Shakin-Eshleman, S. H. Regulation of N-linked core glycosylation: use of a site-directed mutagenesis approach to identify Asn-Xaa-Ser/Thr sequons that are poor oligosacchride acceptors. Biochem. J. 1997, 323, 415-419. Mellquist, J. L.; Kasturi, L.; Spitalnik, S. L.; Shakin-Eshleman, S. H. The amino acid following an Asn-X-Ser/Thr sequon is an important determinant of N-linked core glycosylation efficiency. Biochemistry 1998, 37, 6833-6837). Further simulations on purely theoretical sequences suggested that influences of charged residues were a subset of multiple mechanisms in the determination of glycosylation site-occupancy.

Neural-network-based identification of tissue-type plasminogen activator protein production and glycosylation in CHO cell culture under shear environment.

An artificial neural network (ANN) modeling scheme has been constructed for the identification of both recombinant tissue-type plasminogen activator (r-tPA) protein production and glycosylation from Chinese hamster ovary (CHO) cell culture, cultivated in a stirred bioreactor. A series of hybrid feed-forward backpropagation neural networks were constructed to function as a software sensor. This enabled predictions of viable cell density, r-tPA content, and r-tPA glycosylation. The sensor was based on an initial input vector space consisting of simple metabolite concentrations, batch cultivation time, and a description of shear stress applied to the culture. Metabolite concentrations of the culture supernatant, included in the input vector space, were obtained from a single isocratic HPLC measurement. The shear stress component of the input space enabled accurate culture state prediction over a wide range of agitation rates. Coefficient of determination (r(2)) values between ANN predicted and experimental measurements of 0.945, 0.943, 0.956, and 0.990 were calculated to validate individual ANN prediction accuracy for total ammonia, apparent viable cell density, total r-tPA, and Type II glycoform concentrations, respectively.