Amino acid content of recombinant proteins influences the metabolic burden response

Toward Microbiome Engineering: Expanding the Repertoire of Genetically Tractable Members of the Human Gut Microbiome

Genetic manipulation is necessary to interrogate the functions of microbes in their environments, such as the human gut microbiome. Yet, the vast majority of human gut microbiome species are not genetically tractable. Here, we review the hurdles to seizing genetic control of more species. We address the barriers preventing the application of genetic techniques to gut microbes and report on genetic systems currently under development. While methods aimed at genetically transforming many species simultaneously in situ show promise, they are unable to overcome many of the same challenges that exist for individual microbes. Unless a major conceptual breakthrough emerges, the genetic tractability of the microbiome will remain an arduous task. Increasing the list of genetically tractable organisms from the human gut remains one of the highest priorities for microbiome research and will provide the foundation for microbiome engineering.

PERISCOPE-Opt: Machine learning-based prediction of optimal fermentation conditions and yields of recombinant periplasmic protein expressed in Escherichia coli

•

The ensemble model considered both fermentation conditions and protein properties.

•

Optimal fermentation conditions and periplasmic recombinant protein yield can be predicted.

•

Predictor’s accuracy and Pearson correlation coefficient are 75% and 0.91, respectively.

Optimization of the fermentation process for recombinant protein production (RPP) is often resource-intensive. Machine learning (ML) approaches are helpful in minimizing the experimentations and find vast applications in RPP. However, these ML-based tools primarily focus on features with respect to amino-acid-sequence, ruling out the influence of fermentation process conditions. The present study combines the features derived from fermentation process conditions with that from amino acid-sequence to construct an ML-based model that predicts the maximal protein yields and the corresponding fermentation conditions for the expression of target recombinant protein in the Escherichia coli periplasm. Two sets of XGBoost classifiers were employed in the first stage to classify the expression levels of the target protein as high (>50 mg/L), medium (between 0.5 and 50 mg/L), or low (<0.5 mg/L). The second-stage framework consisted of three regression models involving support vector machines and random forest to predict the expression yields corresponding to each expression-level-class. Independent tests showed that the predictor achieved an overall average accuracy of 75% and a Pearson coefficient correlation of 0.91 for the correctly classified instances. Therefore, our model offers a reliable substitution of numerous trial-and-error experiments to identify the optimal fermentation conditions and yield for RPP. It is also implemented as an open-access webserver, PERISCOPE-Opt (http://periscope-opt.erc.monash.edu).

Just the Two of Us? A Family of Pseudomonas Megaplasmids Offers a Rare Glimpse into the Evolution of Large Mobile Elements

Pseudomonads are ubiquitous group of environmental proteobacteria, well known for their roles in biogeochemical cycling, in the breakdown of xenobiotic materials, as plant growth promoters, and as pathogens of a variety of host organisms. We have previously identified a large megaplasmid present within one isolate of the plant pathogen Pseudomonas syringae, and here we report that a second member of this megaplasmid family is found within an environmental Pseudomonad isolate most closely related to Pseudomonas putida. Many of the shared genes are involved in critical cellular processes like replication, transcription, translation, and DNA repair. We argue that presence of these shared pathways sheds new light on discussions about the types of genes that undergo horizontal gene transfer (i.e., the complexity hypothesis) as well as the evolution of pangenomes. Furthermore, although both megaplasmids display a high level of synteny, genes that are shared differ by over 50% on average at the amino acid level. This combination of conservation in gene order despite divergence in gene sequence suggests that this Pseudomonad megaplasmid family is relatively old, that gene order is under strong selection within this family, and that there are likely many more members of this megaplasmid family waiting to be found in nature.

Scaling up genetic circuit design for cellular computing: advances and prospects

Synthetic biology aims to engineer and redesign biological systems for useful real-world applications in biomanufacturing, biosensing and biotherapy following a typical design-build-test cycle. Inspired from computer science and electronics, synthetic gene circuits have been designed to exhibit control over the flow of information in biological systems. Two types are Boolean logic inspired TRUE or FALSE digital logic and graded analog computation. Key principles for gene circuit engineering include modularity, orthogonality, predictability and reliability. Initial circuits in the field were small and hampered by a lack of modular and orthogonal components, however in recent years the library of available parts has increased vastly. New tools for high throughput DNA assembly and characterization have been developed enabling rapid prototyping, systematic in situ characterization, as well as automated design and assembly of circuits. Recently implemented computing paradigms in circuit memory and distributed computing using cell consortia will also be discussed. Finally, we will examine existing challenges in building predictable large-scale circuits including modularity, context dependency and metabolic burden as well as tools and methods used to resolve them. These new trends and techniques have the potential to accelerate design of larger gene circuits and result in an increase in our basic understanding of circuit and host behaviour.

Integrative analysis of fitness and metabolic effects of plasmids in Pseudomonas aeruginosa PAO1

Horizontal gene transfer (HGT) mediated by the spread of plasmids fuels evolution in prokaryotes. Although plasmids provide bacteria with new adaptive genes, they also produce physiological alterations that often translate into a reduction in bacterial fitness. The fitness costs associated with plasmids represent an important limit to plasmid maintenance in bacterial communities, but their molecular origins remain largely unknown. In this work, we combine phenomics, transcriptomics and metabolomics to study the fitness effects produced by a collection of diverse plasmids in the opportunistic pathogen Pseudomonas aeruginosa PAO1. Using this approach, we scan the physiological changes imposed by plasmids and test the generality of some main mechanisms that have been proposed to explain the cost of HGT, including increased biosynthetic burden, reduced translational efficiency, and impaired chromosomal replication. Our results suggest that the fitness effects of plasmids have a complex origin, since none of these mechanisms could individually provide a general explanation for the cost of plasmid carriage. Interestingly, our results also showed that plasmids alter the expression of a common set of metabolic genes in PAO1, and produce convergent changes in host cell metabolism. These surprising results suggest that there is a common metabolic response to plasmids in P. aeruginosa PAO1.

Fitness Costs of Plasmids: a Limit to Plasmid Transmission

Plasmids mediate the horizontal transmission of genetic information between bacteria, facilitating their adaptation to multiple environmental conditions. An especially important example of the ability of plasmids to catalyze bacterial adaptation and evolution is their instrumental role in the global spread of antibiotic resistance, which constitutes a major threat to public health. Plasmids provide bacteria with new adaptive tools, but they also entail a metabolic burden that, in the absence of selection for plasmid-encoded traits, reduces the competitiveness of the plasmid-carrying clone. Although this fitness reduction can be alleviated over time through compensatory evolution, the initial cost associated with plasmid carriage is the main constraint on the vertical and horizontal replication of these genetic elements. The fitness effects of plasmids therefore have a crucial influence on their ability to associate with new bacterial hosts and consequently on the evolution of plasmid-mediated antibiotic resistance. However, the molecular mechanisms underlying plasmid fitness cost remain poorly understood. Here, we analyze the literature in the field and examine the potential fitness effects produced by plasmids throughout their life cycle in the host bacterium. We also explore the various mechanisms evolved by plasmids and bacteria to minimize the cost entailed by these mobile genetic elements. Finally, we discuss potential future research directions in the field.

MCbiclust: a novel algorithm to discover large-scale functionally related gene sets from massive transcriptomics data collections

The potential to understand fundamental biological processes from gene expression data has grown in parallel with the recent explosion of the size of data collections. However, to exploit this potential, novel analytical methods are required, capable of discovering large co-regulated gene networks. We found current methods limited in the size of correlated gene sets they could discover within biologically heterogeneous data collections, hampering the identification of multi-gene controlled fundamental cellular processes such as energy metabolism, organelle biogenesis and stress responses. Here we describe a novel biclustering algorithm called Massively Correlated Biclustering (MCbiclust) that selects samples and genes from large datasets with maximal correlated gene expression, allowing regulation of complex networks to be examined. The method has been evaluated using synthetic data and applied to large bacterial and cancer cell datasets. We show that the large biclusters discovered, so far elusive to identification by existing techniques, are biologically relevant and thus MCbiclust has great potential in the analysis of transcriptomics data to identify large-scale unknown effects hidden within the data. The identified massive biclusters can be used to develop improved transcriptomics based diagnosis tools for diseases caused by altered gene expression, or used for further network analysis to understand genotype-phenotype correlations.

Et tu, Brute? Not Even Intracellular Mutualistic Symbionts Escape Horizontal Gene Transfer

Many insect species maintain mutualistic relationships with endosymbiotic bacteria. In contrast to their free-living relatives, horizontal gene transfer (HGT) has traditionally been considered rare in long-term endosymbionts. Nevertheless, meta-omics exploration of certain symbiotic models has unveiled an increasing number of bacteria-bacteria and bacteria-host genetic transfers. The abundance and function of transferred loci suggest that HGT might play a major role in the evolution of the corresponding consortia, enhancing their adaptive value or buffering detrimental effects derived from the reductive evolution of endosymbionts’ genomes. Here, we comprehensively review the HGT cases recorded to date in insect-bacteria mutualistic consortia, and discuss their impact on the evolutionary success of these associations.

The relationship of metabolic burden to productivity levels in CHO cell lines

The growing demand for recombinant therapeutics has driven biotechnologists to develop new production strategies. One such strategy for increasing the expression of heterologous proteins has focused on enhancing cell-specific productivity through environmental perturbations. In this work, the effects of hypothermia, hyperosmolarity, high shear stress, and sodium butyrate treatment on growth and productivity were studied using three (low, medium, and high producing) CHO cell lines that differed in their specific productivities of monoclonal antibody. In all three cell lines, the inhibitory effect of these parameters on proliferation was demonstrated. Additionally, compared to the control, specific productivity was enhanced under all conditions and exhibited a consistent cell line specific pattern, with maximum increases (50-290%) in the low producer, and minimum increases (7-20%) in the high producer. Thus, the high-producing cell line was less responsive to environmental perturbations than the low-producing cell line. We hypothesize that this difference is most likely due to the bottleneck associated with a higher metabolic burden caused by higher antibody expression. Increased recombinant mRNA levels and pyruvate carboxylase activities due to low temperature and hyperosmotic stress were found to be positively associated with the metabolic burden.

Prediction of Escherichia coli expression performance in microtiter plates by analyzing only the temporal development of scattered light during culture

Background

Escherichia coli is often used for recombinant protein production. The expression of recombinant proteins negatively affects the microbial growth, thus, a balance between protein expression and biomass formation is preferable to reach high product- and space-time-yield. Already in screening experiments, suboptimal conditions causing too weak or too strong induction must be avoided. High-throughput screening devices such as the BioLector are often applied for screening experiments. The BioLector allows optical online monitoring of each well in a continuously orbitally shaken microtiter plate via scattered light and fluorescence measurements. This technique enables a fast identification of promising clones. However, to determine the expression performance of non-fluorescent products elaborated offline analysis is often required.

Methods

A mathematical method is developed to distinguish between cultures, which are insufficiently, optimally or too strongly induced. Therefore, just the temporal development of the scattered light intensity signal is investigated. It is found that discrimination between the different intensities of induction is possible via principal component analysis. By fitting an extended sigmoidal function to the trajectory of the scattered light over time, two characteristic parameters are found. These are used in an empirical model to predict the expression performance.

Results

The method was established for a wide range of culture conditions based on 625 E. coli cultures. Three E. coli host strains (Tuner(DE3), BL21(DE3), and BL21-Gold(DE3)) expressing either flavin-mononucleotide-based fluorescent protein (FbFP) or Cellulase celA2 were investigated. Cultures were conducted in two different types of microtiter plates (48- and 96-wells), in two online measurement devices at four temperatures (28 °C, 30 °C, 34 °C, and 37 °C). More than 95% of the predicted values are in agreement with the offline measured expression performances with a satisfying accuracy of ±30%.

Conclusions

The properties of cultures studied can be represented by only two characteristic parameters (slope at and time of the inflection point) received from fitting an extended sigmoidal function to the respective scattered light trajectory. Based on these two characteristic parameters, predictions of the standardized expression performance are possible and for a first screen elaborated offline analysis can be avoided. To the best of our knowledge, this is the first work presenting a method for the general prediction of expression performance of E. coli based solely on the temporal development of scattered light signals.

Electronic supplementary material

The online version of this article (doi:10.1186/s13036-017-0064-5) contains supplementary material, which is available to authorized users.

Synthetic gene design-The rationale for codon optimization and implications for molecular pharming in plants

Degeneracy in the genetic code allows multiple codon sequences to encode the same protein. Codon usage bias in genes is the term given to the preferred use of particular synonymous codons. Synonymous codon substitutions had been regarded as "silent" as the primary structure of the protein was not affected; however, it is now accepted that synonymous substitutions can have a significant effect on heterologous protein expression. Codon optimization, the process of altering codons within the gene sequence to improve recombinant protein expression, has become widely practised. Multiple inter-linked factors affecting protein expression need to be taken into consideration when optimizing a gene sequence. Over the years, various computer programmes have been developed to aid in the gene sequence optimization process. However, as the rulebook for altering codon usage to affect protein expression is still not completely understood, it is difficult to predict which strategy, if any, will design the "optimal" gene sequence. In this review, codon usage bias and factors affecting codon selection will be discussed and the evidence for codon optimization impact will be reviewed for recombinant protein expression using plants as a case study. These developments will be relevant to all recombinant expression systems; however, molecular pharming in plants is an area which has consistently encountered difficulties with low levels of recombinant protein expression, and should benefit from an evidence based rational approach to synthetic gene design. Biotechnol. Bioeng. 2017;114: 492-502. © 2016 Wiley Periodicals, Inc.

A particular silent codon exchange in a recombinant gene greatly influences host cell metabolic activity

BACKGROUND

Recombinant protein production using Escherichia coli as expression host is highly efficient, however, it also induces strong host cell metabolic burden. Energy and biomass precursors are withdrawn from the host's metabolism as they are required for plasmid replication, heterologous gene expression and protein production. Rare codons in a heterologous gene may be a further drawback. This study aims to investigate the influence of particular silent codon exchanges within a heterologous gene on host cell metabolic activity. Silent mutations were introduced into the coding sequence of a model protein to introduce all synonymous arginine or leucine codons at two randomly defined positions, as well as substitutions leading to identical amino acid exchanges with different synonymous codons. The respective E. coli clones were compared during cultivation in a mineral autoinduction medium using specialized online and offline measuring techniques to quantitatively analyze effects on respiration, biomass and protein production, as well as on carbon source consumption, plasmid copy number, intracellular nucleobases and mRNA content of each clone.

RESULTS

Host stain metabolic burden correlates with recombinant protein production. Upon heterologous gene expression, tremendous differences in respiration, biomass and protein production were observed. According to their different respiration activity the E. coli clones could be classified into two groups, Type A and Type B. Type A clones tended to higher product formation, Type B clones showed stronger biomass formation. Whereas codon usage and intracellular nucleobases had no influence on the Type-A-Type-B-behavior, plasmid copy number, mRNA content and carbon source consumption strongly differed between the two groups.

CONCLUSIONS

Particular silent codon exchanges in a heterologous gene sequence led to differences in initial growth of Type A and Type B clones. Thus, the biomass concentration at the time point of induction varied. In consequence, not only plasmid copy number and expression levels differed between the two groups, but also the kinetics of lactose and glycerol consumption. Even though the underlying molecular mechanisms are not yet identified we observed the astonishing phenomenon that particular silent codon exchanges within a heterologous gene tremendously affect host cell metabolism and recombinant protein production. This could have great impact on codon optimization of heterologous genes, screening procedures for improved variants, and biotechnological protein production processes.

Quantifying cellular capacity identifies gene expression designs with reduced burden

Heterologous gene expression can be a significant burden for cells. Here we describe an in vivo monitor that tracks changes in the capacity of Escherichia coli in real time and can be used to assay the burden imposed by synthetic constructs and their parts. We identify construct designs with reduced burden that predictably outperformed less efficient designs, despite having equivalent output.

Exchange of single amino acids at different positions of a recombinant protein affects metabolic burden in Escherichia coli

Background

Escherichia coli is commonly used in academia and industry for expressing recombinant proteins because of its well-characterized molecular genetics and the availability of numerous expression vectors and strains. One important issue during recombinant protein production is the so-called ‘metabolic burden’: the material and energy normally reserved for microbial metabolism which is sapped from the bacterium to produce the recombinant protein. This material and energy drain harms biomass formation and modifies respiration. To the best of our knowledge, no research has investigated so far whether a single amino acid exchange in a recombinant protein affects the metabolic burden phenomenon. Thus, in this study, 15 E. coli BL21(DE3) clones expressing either the fusion tags, a recombinant wild type lipase, or 13 different lipase variants are investigated to quantitatively analyze the respective effects of single amino acid exchanges at different positions on respiration, biomass and protein production of each clone. Therefore, two small-scale online monitoring systems, namely a Respiration Activity MOnitoring System (RAMOS) and a microtiter plate based cultivation system (BioLector) are applied.

Results

Upon expression of all enzyme variants, strong variations were found in the Oxygen Transfer Rate (OTR), biomass and protein (lipase) production of the respective E. coli clones. Two distinct patterns of respiration behavior were observed and, so, the clones could be classified into two groups (Type A and B). Potential factors to explain these patterns were evaluated (e.g. plasmid copy number, inclusion body formation). However, no decisive factor could yet be identified. Five distinct cultivation phases could be determined from OTR curves which give real-time information about carbon source consumption, biomass and protein production. In general, it was found that the quantity of product increased with the duration of active respiration.

Conclusions

This work demonstrates that single amino acid exchanges in a recombinant protein influence the metabolic burden during protein production. The small-scale online monitoring devices RAMOS and BioLector enable the real-time detection of even smallest differences in respiration behavior, biomass and protein production in the E. coli clones investigated. Hence, this study underscores the importance of parallel online monitoring systems to unveil the relevance of single amino acid exchanges for the recombinant protein production.

Electronic supplementary material

The online version of this article (doi:10.1186/s12934-015-0191-y) contains supplementary material, which is available to authorized users.

Genome engineering for improved recombinant protein expression in Escherichia coli

A metabolic engineering perspective which views recombinant protein expression as a multistep pathway allows us to move beyond vector design and identify the downstream rate limiting steps in expression. In E.coli these are typically at the translational level and the supply of precursors in the form of energy, amino acids and nucleotides. Further recombinant protein production triggers a global cellular stress response which feedback inhibits both growth and product formation. Countering this requires a system level analysis followed by a rational host cell engineering to sustain expression for longer time periods. Another strategy to increase protein yields could be to divert the metabolic flux away from biomass formation and towards recombinant protein production. This would require a growth stoppage mechanism which does not affect the metabolic activity of the cell or the transcriptional or translational efficiencies. Finally cells have to be designed for efficient export to prevent buildup of proteins inside the cytoplasm and also simplify downstream processing. The rational and the high throughput strategies that can be used for the construction of such improved host cell platforms for recombinant protein expression is the focus of this review.

Influence of hydrophilic amino acids and GC-content on expression of recombinant proteins used in vaccines against foot-and-mouth disease virus in Escherichia coli

Epitope-based protein expression in Escherichia coli can be improved by adjusting its amino acid composition and encoding genes. To that end, we analyzed 24 recombinant epitope proteins (rEPs) that carry multiple epitopes derived from VP1 protein of foot-and-mouth disease virus. High level expression of the rEPs was attributed to a high content of Arg, Asn, Asp and Thr, a low content of Gln, Pro and Lys, a high content of hydrophilic amino acids and a higher isoelectric point value resulting from abundant Arg. It is also attributed to the appropriate guanine and cytosine content in the encoding genes. The data provide a reference for adjusting the amino acid composition in designing epitope-based proteins used in vaccines and for adjusting the synonymous codons to improve their expressions in E. coli.

Transient growth arrest in Escherichia coli induced by chromosome condensation

MukB is a bacterial SMC (structural maintenance of chromosome) protein that regulates the global folding of the Escherichia coli chromosome by bringing distant DNA segments together. We report that moderate overproduction of MukB may lead, depending on strain and growth conditions, to transient growth arrest. In DH5α cells, overproduction of MukB or MukBEF using pBAD expression system triggered growth arrest 2.5 h after induction. The exit from growth arrest was accompanied by the loss of the overproducing plasmid and a decline in the abundance of MukBEF. The arrested cells showed a compound gene expression profile which can be characterized by the following features: (i) a broad and deep downregulation of ribosomal proteins (up to 80-fold); (ii) downregulation of groups of genes encoding enzymes involved in nucleotide metabolism, respiration, and central metabolism; (iii) upregulation of some of the genes responsive to general stress; and (iv) degradation of the patterns of spatial correlations in the transcriptional activity of the chromosome. The transcriptional state of the MukB induced arrest is most similar to stationary cells and cells recovered from stationary phase into a nutrient deprived medium, to amino acid starved cells and to the cells shifting from glucose to acetate. The mukB++ state is dissimilar from all examined transcriptional states generated by protein overexpression with the possible exception of RpoE and RpoH overexpression. Thus, the transcription profile of MukB-arrested cells can be described as a combination of responses typical for other growth-arrested cells and those for overproducers of DNA binding proteins with a particularly deep down-regulation of ribosomal genes.

Translational sensitivity of the Escherichia coli genome to fluctuating tRNA availability

The synthesis of protein from messenger RNA during translation is a highly dynamic process that plays a key role in controlling the efficiency and fidelity of genome-wide protein expression. The availability of aminoacylated transfer RNA (tRNA) is a major factor influencing the speed of ribosomal movement, which depending on codon choices, varies considerably along a transcript. Furthermore, it has been shown experimentally that tRNA availability can vary significantly under different growth and stress conditions, offering the cell a way to adapt translational dynamics across the genome. Existing models of translation have neglected fluctuations of tRNA pools, instead assuming fixed tRNA availabilities over time. This has lead to an incomplete understanding of this process. Here, we show for the entire Escherichia coli genome how and to what extent translational speed profiles, which capture local aspects of translational elongation, respond to measured shifts in tRNA availability. We find that translational profiles across the genome are affected to differing degrees, with genes that are essential or related to fundamental processes such as translation, being more robust than those linked to regulation. Furthermore, we reveal how fluctuating tRNA availability influences profiles of specific sequences known to play a significant role in translational control of gene expression.

Taxonomy and evolution of bacteriochlorophyll a-containing members of the OM60/NOR5 clade of marine gammaproteobacteria: description of Luminiphilus syltensis gen. nov., sp. nov., reclassification of Haliea rubra as Pseudohaliea rubra gen. nov., comb. nov., and emendation of Chromatocurvus halotolerans

Background

Aerobic gammaproteobacteria affiliated to the OM60/NOR5 clade are widespread in saline environments and of ecological importance in several marine ecosystems, especially the euphotic zone of coastal areas. Within this group a close relationship between aerobic anoxygenic photoheterotrophs and non-phototrophic members has been found.

Results

Several strains of aerobic red-pigmented bacteria affiliated to the OM60/NOR5 clade were obtained from tidal flat sediment samples at the island of Sylt (North Sea, Germany). Two of the novel isolates, Rap1red and Ivo14^T, were chosen for an analysis in detail. Strain Rap1red shared a 16S rRNA sequence identity of 99% with the type strain of Congregibacter litoralis and was genome-sequenced to reveal the extent of genetic microheterogeneity among closely related strains within this clade. In addition, a draft genome sequence was obtained from the isolate Ivo14^T, which belongs to the environmental important NOR5-1 lineage that contains so far no cultured representative with a comprehensive description. Strain Ivo14^T was characterized using a polyphasic approach and compared with other red-pigmented members of the OM60/NOR5 clade, including Congregibacter litoralis DSM 17192^T, Haliea rubra DSM 19751^T and Chromatocurvus halotolerans DSM 23344^T. All analyzed strains contained bacteriochlorophyll a and spirilloxanthin as photosynthetic pigments. Besides a detailed phenotypic characterization including physiological and chemotaxonomic traits, sequence information based on protein-coding genes and a comparison of draft genome data sets were used to identify possible features characteristic for distinct taxa within this clade.

Conclusions

Comparative sequence analyses of the pufLM genes of genome-sequenced representatives of the OM60/NOR5 clade indicated that the photosynthetic apparatus of these species was derived from a common ancestor and not acquired by multiple horizontal gene transfer from phylogenetically distant species. An affiliation of the characterized bacteriochlorophyll a-containing strains to different genera was indicated by significant phenotypic differences and pufLM nucleotide sequence identity values below 82%. The revealed high genotypic and phenotypic diversity of closely related strains within this phylogenetic group reflects a rapid evolution and frequent niche separation in the OM60/NOR5 clade, which is possibly driven by the necessities of an adaptation to oligotrophic marine habitats.

Enhancing toxic protein expression in Escherichia coli fed-batch culture using kinetic parameters: Human granulocyte-macrophage colony-stimulating factor as a model system

The kinetics of recombinant human granulocyte-macrophage colony-stimulating factor (hGM-CSF) expression was studied under the strong T7 promoter in continuous culture of Escherichia coli using complex medium to design an optimum feeding strategy for high cell density cultivation. Continuous culture studies were done at different dilution rates and the growth and product formation profiles were monitored post-induction. Recombinant protein expression was in the form of inclusion bodies with a maximum specific product formation rate (q(p)) of 63.5 mg g(-1) DCW h(-1) at a dilution rate (D) of 0.3 h(-1). The maximum volumetric product concentration achieved at this dilution rate was 474 mg l(-1), which translated a ~1.4 and ~1.75 folds increase than the values obtained at dilution rates of 0.2 h(-1) and 0.4 h(-1) respectively. The specific product yield (Y(P/x)) peaked at 138 mg g(-1) DCW, demonstrating a ~1.6 folds increase in the values obtained at other dilution rates. A drop in q(p) was observed within 5-6 h of induction at all the dilution rates, possibly due to protein toxicity and metabolic stress associated with protein expression. The data from the continuous culture studies allowed us to design an optimal feeding strategy and induction time in fed-batch cultures which resulted in a maximum product concentration of 3.95 g l(-1) with a specific hGM-CSF yield (Y(P/x)) of 107 mg g(-1) DCW.

Metabolic responses to recombinant bioprocesses in Escherichia coli

Escherichia coli has been widely used for the production of recombinant proteins. However, the unbalances between host metabolism and recombinant biosynthesis continue to hamper the efficiency of these recombinant bioprocesses. The additional drainage of biosynthetic precursors toward recombinant processes burdens severely the metabolism of cells that, ultimately, elicits a series of stress responses, reducing biomass growth and recombinant protein production. Several strategies to overcome these metabolic limitations have been implemented; however, in most cases, improvements in recombinant protein expression were achieved at the expense of biomass growth arrest, which significantly hampers the efficiency of recombinant bioprocesses. With the advent of high throughput techniques and modelling approaches that provide a system-level understanding of the cellular systems, it is now expected that new advances in recombinant bioprocesses are achieved. By providing means to deal with these systems, our understanding on the metabolic behaviour of recombinant cells will advance and can be further explored to the design of suitable hosts and more efficient and cost-effective bioprocesses. Here, we review the major metabolic responses associated with recombinant processes and the engineering strategies relevant to overcome these stresses. Moreover, the advantages of applying systems levels engineering strategies to enhance recombinant protein production in E. coli cells are discussed and future perspectives on the advances of mathematical modelling approaches to study these systems are exposed.

Engineering genes for predictable protein expression

The DNA sequence used to encode a polypeptide can have dramatic effects on its expression. Lack of readily available tools has until recently inhibited meaningful experimental investigation of this phenomenon. Advances in synthetic biology and the application of modern engineering approaches now provide the tools for systematic analysis of the sequence variables affecting heterologous expression of recombinant proteins. We here discuss how these new tools are being applied and how they circumvent the constraints of previous approaches, highlighting some of the surprising and promising results emerging from the developing field of gene engineering.

Analyzing the metabolic stress response of recombinant Escherichia coli cultures expressing human interferon-beta in high cell density fed batch cultures using time course transcriptomic data

Fed batch cultures expressing recombinant interferon beta under the T7 promoter were run with different exponential feeding rates of a complex substrate and induced at varying cell densities. Post-induction profiles of the specific product formation rates showed a strong dependence on the specific growth rate with the maximum product yield obtained at 0.2 h(-1). A study of the relative transcriptomic profiles as a function of pre-induction μ was therefore done to provide insight into the role of cellular physiology in enhancing recombinant protein expression. Hierarchical clustering analysis of the significantly regulated genes allowed us to identify biologically important groups of genes which fall under specific master regulators. The groups were: rpoH, ArcB, CreB, Lrp, RelA, Fis and Hfq. The response of these regulators, which exert a feedback control on the growth and product formation rates correlated well with the expression levels obtained. Thus at the optimum pre-induction μ, the alternative sigma factors and ribosomal machinery genes did not get depressed till the 6th hour post-induction unlike at other specific growth rates, demonstrating a critical role for the genes in sustaining recombinant protein expression.

High yields of 2,3-butanediol and mannitol in Lactococcus lactis through engineering of NAD⁺ cofactor recycling

Manipulation of NADH-dependent steps, and particularly disruption of the las-located lactate dehydrogenase (ldh) gene in Lactococcus lactis, is common to engineering strategies envisaging the accumulation of reduced end products other than lactate. Reverse transcription-PCR experiments revealed that three out of the four genes assigned to lactate dehydrogenase in the genome of L. lactis, i.e., the ldh, ldhB, and ldhX genes, were expressed in the parental strain MG1363. Given that genetic redundancy is often a major cause of metabolic instability in engineered strains, we set out to develop a genetically stable lactococcal host tuned for the production of reduced compounds. Therefore, the ldhB and ldhX genes were sequentially deleted in L. lactis FI10089, a strain with a deletion of the ldh gene. The single, double, and triple mutants, FI10089, FI10089ΔldhB, and FI10089ΔldhBΔldhX, showed similar growth profiles and displayed mixed-acid fermentation, ethanol being the main reduced end product. Hence, the alcohol dehydrogenase-encoding gene, the adhE gene, was inactivated in FI10089, but the resulting strain reverted to homolactic fermentation due to induction of the ldhB gene. The three lactate dehydrogenase-deficient mutants were selected as a background for the production of mannitol and 2,3-butanediol. Pathways for the biosynthesis of these compounds were overexpressed under the control of a nisin promoter, and the constructs were analyzed with respect to growth parameters and product yields under anaerobiosis. Glucose was efficiently channeled to mannitol (maximal yield, 42%) or to 2,3-butanediol (maximal yield, 67%). The theoretical yield for 2,3-butanediol was achieved. We show that FI10089ΔldhB is a valuable basis for engineering strategies aiming at the production of reduced compounds.

Model based optimization of the fed-batch production of a highly active transglutaminase variant in Escherichia coli

A process for the production of a thermostable variant of a microbial transglutaminase was developed. The transglutaminase variant produced, carried a single amino acid exchange (serine replaced by proline at position 2) and showed a nearly doubled specific activity of 46.1 Umg(-1) compared to the wild-type enzyme. Based on a model based optimization strategy, intracellular soluble production in Escherichia coli was optimized. After parameter identification and only two fed-batch cultivations, a space time yield of 1438 U(TG)L(-1)h(-1) was obtained which is 175% higher than the highest values published so far (extracellular production using Corynebacterium ammoniagenes). High carbon source concentrations during expression were found to increase the product formation. Prior to the fed-batch cultivation, the host strain was adapted from complex medium to minimal medium by serial dilution. Upon transfer to the minimal medium, initially the maximal growth rate dropped to 0.13 h(-1). After the six consecutive cultivations the rate increased to 0.47 h(-1) and the portion of the complex medium was reduced to 1 ppm. Using the adapted cells, temperature after induction and IPTG-concentration were investigated by satellite batch cultivation according to a Design of Experiment (DoE) plan. The product yield was strongly influenced by the temperature after induction but not by the inductor concentration. The highest specific activity of 1386 Ug(-1) bio dry mass was obtained at 29°C and 0.7 mM IPTG.

Physiological relation between respiration activity and heterologous expression of selected benzoylformate decarboxylase variants in Escherichia coli

BACKGROUND

The benzoylformate decarboxylase (BFD) from Pseudomonas putida is a biotechnologically interesting biocatalyst. It catalyses the formation of chiral 2-hydroxy ketones, which are important building blocks for stereoselective syntheses. To optimise the enzyme function often the amino acid composition is modified to improve the performance of the enzyme. So far it was assumed that a relatively small modification of the amino acid composition of a protein does not significantly influence the level of expression or media requirements. To determine, which effects these modifications might have on cultivation and product formation, six different BFD-variants with one or two altered amino acids and the wild type BFD were expressed in Escherichia coli SG13009 pKK233-2. The oxygen transfer rate (OTR) as parameter for growth and metabolic activity of the different E. coli clones was monitored on-line in LB, TB and modified PanG mineral medium with the Respiratory Activity MOnitoring System (RAMOS).

RESULTS

Although the E. coli clones were genetically nearly identical, the kinetics of their metabolic activity surprisingly differed in the standard media applied. Three different types of OTR curves could be distinguished. Whereas the first type (clones expressing Leu476Pro-Ser181Thr or Leu476Pro) had typical OTR curves, the second type (clones expressing the wild type BFD, Ser181Thr or His281Ala) showed an early drop of OTR in LB and TB medium and a drastically reduced maximum OTR in modified PanG mineral medium. The third type (clone expressing Leu476Gln) behaved variable. Depending on the cultivation conditions, its OTR curve was similar to the first or the second type. It was shown, that the kinetics of the metabolic activity of the first type depended on the concentration of thiamine, which is a cofactor of BFD, in the medium. It was demonstrated that the cofactor binding strength of the different BFD-variants correlated with the differences in metabolic activity of their respective host strain.

CONCLUSIONS

The BFD-variants with high cofactor binding affinity (wild type, His281Ala, Ser181Thr) obviously extract thiamine from the medium and bind it tightly to the enzyme. This might explain the hampered growth of these clones. In contrast, growth of clones expressing variants with low cofactor binding affinity (Leu476His, Leu476Pro, Leu476Pro-Ser181Thr) is not impaired. Leu476Gln has an intermediate cofactor binding strength, thus, growth of its host strain depends on the specific cultivation conditions. This paper shows that slight differences of the amino acid composition can affect protein expression and cultivation and might require an adaptation of media components. Effects such as the observed are hardly foreseeable and difficult to detect in conventional screening processes. Via small scale experiments with on-line measurements in shake flasks such effects influencing the cultivation and product formation can be detected and avoided.

Dissecting the expression patterns of transcription factors across conditions using an integrated network-based approach

In prokaryotes, regulation of gene expression is predominantly controlled at the level of transcription. Transcription in turn is mediated by a set of DNA-binding factors called transcription factors (TFs). In this study, we map the complete repertoire of ∼300 TFs of the bacterial model, Escherichia coli, onto gene expression data for a number of nonredundant experimental conditions and show that TFs are generally expressed at a lower level than other gene classes. We also demonstrate that different conditions harbor varying number of active TFs, with an average of about 15% of the total repertoire, with certain stress and drug-induced conditions exhibiting as high as one-third of the collection of TFs. Our results also show that activators are more frequently expressed than repressors, indicating that activation of promoters might be a more common phenomenon than repression in bacteria. Finally, to understand the association of TFs with different conditions and to elucidate their dynamic interplay with other TFs, we develop a network-based framework to identify TFs which act as markers, defined as those which are responsible for condition-specific transcriptional rewiring. This approach allowed us to pinpoint several marker TFs as being central in various specialized conditions such as drug induction or growth condition variations, which we discuss in light of previously reported experimental findings. Further analysis showed that a majority of identified markers effectively control the expression of their regulons and, in general, transcriptional programs of most conditions can be effectively rewired by a very small number of TFs. It was also found that closeness is a key centrality measure which can aid in the successful identification of marker TFs in regulatory networks. Our results suggest the utility of the network-based approaches developed in this study to be applicable for understanding other interactomic data sets.

Increased mutation frequency in redox-impaired Escherichia coli due to RelA- and RpoS-mediated repression of DNA repair

Balancing of reducing equivalents is a fundamental issue in bacterial metabolism and metabolic engineering. Mutations in the key metabolic genes ldhA and pflB of Escherichia coli are known to stall anaerobic growth and fermentation due to a buildup of intracellular NADH. We observed that the rate of spontaneous mutation in E. coli BW25113 (DeltaldhA DeltapflB) was an order of magnitude higher than that in wild-type (WT) E. coli BW25113. We hypothesized that the increased mutation frequency was due to an increased NADH/NAD(+) ratio in this strain. Using several redox-impaired strains of E. coli and different redox conditions, we confirmed a significant correlation (P < 0.01) between intracellular-NADH/NAD(+) ratio and mutation frequency. To identify the genetic basis for this relationship, whole-genome transcriptional profiles were compared between BW25113 WT and BW25113 (DeltaldhA DeltapflB). This analysis revealed that the genes involved in DNA repair were expressed at significantly lower levels in BW25113 (DeltaldhA DeltapflB). Direct measurements of the extent of DNA repair in BW25113 (DeltaldhA DeltapflB) subjected to UV exposure confirmed that DNA repair was inhibited. To identify a direct link between DNA repair and intracellular-redox ratio, the stringent-response-regulatory gene relA and the global-stress-response-regulatory gene rpoS were deleted. In both cases, the mutation frequencies were restored to BW25113 WT levels.

Replication methods and tools in high-throughput cultivation processes - recognizing potential variations of growth and product formation by on-line monitoring

Background

High-throughput cultivations in microtiter plates are the method of choice to express proteins from recombinant clone libraries. Such processes typically include several steps, whereby some of them are linked by replication steps: transformation, plating, colony picking, preculture, main culture and induction. In this study, the effects of conventional replication methods and replication tools (8-channel pipette, 96-pin replicators: steel replicator with fixed or spring-loaded pins, plastic replicator with fixed pins) on growth kinetics of Escherichia coli SCS1 pQE-30 pSE111 were observed. Growth was monitored with the BioLector, an on-line monitoring technique for microtiter plates. Furthermore, the influence of these effects on product formation of Escherichia coli pRhotHi-2-EcFbFP was investigated. Finally, a high-throughput cultivation process was simulated with Corynebacterium glutamicum pEKEx2-phoD-GFP, beginning at the colony picking step.

Results

Applying different replication tools and methods for one single strain resulted in high time differences of growth of the slowest and fastest growing culture. The shortest time difference (0.3 h) was evaluated for the 96 cultures that were transferred with an 8-channel pipette from a thawed and mixed cryoculture and the longest time difference (6.9 h) for cultures that were transferred with a steel replicator with fixed pins from a frozen cryoculture. The on-line monitoring of a simulated high-throughput cultivation process revealed strong variances in growth kinetics and a twofold difference in product formation. Another experiment showed that varying growth kinetics, caused by varying initial biomass concentrations (OD₆₀₀ of 0.0125 to 0.2) led to strongly varying product formation upon induction at a defined point of time.

Conclusions

To improve the reproducibility of high-throughput cultivation processes and the comparability between different applied cultures, it is strongly recommended to use automated or manual liquid handling stations or, alternatively, multi-channel pipettes. Because of their higher transfer volume and hence precision in comparison to pin replicators, they reduce the variance of initial biomass concentrations. With respect to the results obtained, other methods to increase the comparability between parallel cultivations by compensating differences in biomass concentrations are required, such as using autoinduction media, fed-batch operation of precultures or on-line monitoring in microtiter plates combined with automated liquid handling.

Design parameters to control synthetic gene expression in Escherichia coli

BACKGROUND

Production of proteins as therapeutic agents, research reagents and molecular tools frequently depends on expression in heterologous hosts. Synthetic genes are increasingly used for protein production because sequence information is easier to obtain than the corresponding physical DNA. Protein-coding sequences are commonly re-designed to enhance expression, but there are no experimentally supported design principles.

PRINCIPAL FINDINGS

To identify sequence features that affect protein expression we synthesized and expressed in E. coli two sets of 40 genes encoding two commercially valuable proteins, a DNA polymerase and a single chain antibody. Genes differing only in synonymous codon usage expressed protein at levels ranging from undetectable to 30% of cellular protein. Using partial least squares regression we tested the correlation of protein production levels with parameters that have been reported to affect expression. We found that the amount of protein produced in E. coli was strongly dependent on the codons used to encode a subset of amino acids. Favorable codons were predominantly those read by tRNAs that are most highly charged during amino acid starvation, not codons that are most abundant in highly expressed E. coli proteins. Finally we confirmed the validity of our models by designing, synthesizing and testing new genes using codon biases predicted to perform well.

CONCLUSION

The systematic analysis of gene design parameters shown in this study has allowed us to identify codon usage within a gene as a critical determinant of achievable protein expression levels in E. coli. We propose a biochemical basis for this, as well as design algorithms to ensure high protein production from synthetic genes. Replication of this methodology should allow similar design algorithms to be empirically derived for any expression system.

Robo-Lector - a novel platform for automated high-throughput cultivations in microtiter plates with high information content

Background

In industry and academic research, there is an increasing demand for flexible automated microfermentation platforms with advanced sensing technology. However, up to now, conventional platforms cannot generate continuous data in high-throughput cultivations, in particular for monitoring biomass and fluorescent proteins. Furthermore, microfermentation platforms are needed that can easily combine cost-effective, disposable microbioreactors with downstream processing and analytical assays.

Results

To meet this demand, a novel automated microfermentation platform consisting of a BioLector and a liquid-handling robot (Robo-Lector) was sucessfully built and tested. The BioLector provides a cultivation system that is able to permanently monitor microbial growth and the fluorescence of reporter proteins under defined conditions in microtiter plates. Three examplary methods were programed on the Robo-Lector platform to study in detail high-throughput cultivation processes and especially recombinant protein expression. The host/vector system E. coli BL21(DE3) pRhotHi-2-EcFbFP, expressing the fluorescence protein EcFbFP, was hereby investigated. With the method 'induction profiling' it was possible to conduct 96 different induction experiments (varying inducer concentrations from 0 to 1.5 mM IPTG at 8 different induction times) simultaneously in an automated way. The method 'biomass-specific induction' allowed to automatically induce cultures with different growth kinetics in a microtiter plate at the same biomass concentration, which resulted in a relative standard deviation of the EcFbFP production of only ± 7%. The third method 'biomass-specific replication' enabled to generate equal initial biomass concentrations in main cultures from precultures with different growth kinetics. This was realized by automatically transferring an appropiate inoculum volume from the different preculture microtiter wells to respective wells of the main culture plate, where subsequently similar growth kinetics could be obtained.

Conclusion

The Robo-Lector generates extensive kinetic data in high-throughput cultivations, particularly for biomass and fluorescence protein formation. Based on the non-invasive on-line-monitoring signals, actions of the liquid-handling robot can easily be triggered. This interaction between the robot and the BioLector (Robo-Lector) combines high-content data generation with systematic high-throughput experimentation in an automated fashion, offering new possibilities to study biological production systems. The presented platform uses a standard liquid-handling workstation with widespread automation possibilities. Thus, high-throughput cultivations can now be combined with small-scale downstream processing techniques and analytical assays. Ultimately, this novel versatile platform can accelerate and intensify research and development in the field of systems biology as well as modelling and bioprocess optimization.

Metabolic engineering of a reduced-genome strain of Escherichia coli for L-threonine production

Background

Deletion of large blocks of nonessential genes that are not needed for metabolic pathways of interest can reduce the production of unwanted by-products, increase genome stability, and streamline metabolism without physiological compromise. Researchers have recently constructed a reduced-genome Escherichia coli strain MDS42 that lacks 14.3% of its chromosome.

Results

Here we describe the reengineering of the MDS42 genome to increase the production of the essential amino acid L-threonine. To this end, we over-expressed a feedback-resistant threonine operon (thrA*BC), deleted the genes that encode threonine dehydrogenase (tdh) and threonine transporters (tdcC and sstT), and introduced a mutant threonine exporter (rhtA23) in MDS42. The resulting strain, MDS-205, shows an ~83% increase in L-threonine production when cells are grown by flask fermentation, compared to a wild-type E. coli strain MG1655 engineered with the same threonine-specific modifications described above. And transcriptional analysis revealed the effect of the deletion of non-essential genes on the central metabolism and threonine pathways in MDS-205.

Conclusion

This result demonstrates that the elimination of genes unnecessary for cell growth can increase the productivity of an industrial strain, most likely by reducing the metabolic burden and improving the metabolic efficiency of cells.

Genome-scale analysis of anti-metabolite directed strain engineering

Classic strain engineering methods have previously been limited by the low-throughput of conventional sequencing technology. Here, we applied a new genomics technology, scalar analysis of library enrichments (SCALEs), to measure >3 million Escherichia coli genomic library clone enrichment patterns resulting from growth selections employing three aspartic-acid anti-metabolites. Our objective was to assess the extent to which access to genome-scale enrichment patterns would provide strain-engineering insights not reasonably accessible through the use of conventional sequencing. We determined that the SCALEs method identified a surprisingly large range of anti-metabolite tolerance regions (423, 865, or 909 regions for each of the three anti-metabolites) when compared to the number of regions (1-3 regions) indicated by conventional sequencing. Genome-scale methods uniquely enable the calculation of clone fitness values by providing concentration data for all clones within a genomic library before and after a period of selection. We observed that clone fitness values differ substantially from clone concentration values and that this is due to differences in overall clone fitness distributions for each selection. Finally, we show that many of the clones of highest fitness overlapped across all selections, suggesting that inhibition of aspartate metabolism, as opposed to specific inhibited enzymes, dominated each selection. Our follow up studies confirmed our observed growth phenotypes and showed that intracellular amino-acid levels were also altered in several of the identified clones. These results demonstrate that genome-scale methods, such as SCALEs, can be used to dramatically improve understanding of classic strain engineering approaches.

Large-scale mapping and validation of Escherichia coli transcriptional regulation from a compendium of expression profiles

Machine learning approaches offer the potential to systematically identify transcriptional regulatory interactions from a compendium of microarray expression profiles. However, experimental validation of the performance of these methods at the genome scale has remained elusive. Here we assess the global performance of four existing classes of inference algorithms using 445 Escherichia coli Affymetrix arrays and 3,216 known E. coli regulatory interactions from RegulonDB. We also developed and applied the context likelihood of relatedness (CLR) algorithm, a novel extension of the relevance networks class of algorithms. CLR demonstrates an average precision gain of 36% relative to the next-best performing algorithm. At a 60% true positive rate, CLR identifies 1,079 regulatory interactions, of which 338 were in the previously known network and 741 were novel predictions. We tested the predicted interactions for three transcription factors with chromatin immunoprecipitation, confirming 21 novel interactions and verifying our RegulonDB-based performance estimates. CLR also identified a regulatory link providing central metabolic control of iron transport, which we confirmed with real-time quantitative PCR. The compendium of expression data compiled in this study, coupled with RegulonDB, provides a valuable model system for further improvement of network inference algorithms using experimental data.

Balancing a heterologous mevalonate pathway for improved isoprenoid production in Escherichia coli

Engineering biosynthetic pathways in microbes for the production of complex chemicals and pharmaceuticals is an attractive alternative to chemical synthesis. However, in transferring large pathways to alternate hosts and manipulating expression levels, the native regulation of carbon flux through the pathway may be lost leading to imbalances in the pathways. Previously, Escherichia coli was engineered to produce large quantities of isoprenoids by creating a mevalonate-based isopentenyl pyrophosphate biosynthetic pathway [Martin, V.J., Pitera, D.J., Withers, S.T., Newman, J.D., Keasling, J.D., 2003. Engineering a mevalonate pathway in Escherichia coli for production of terpenoids. Nat. Biotechnol. 21, 796-802]. The strain produces high levels of isoprenoids, but upon further investigation we discovered that the accumulation of pathway intermediates limited flux and that high-level expression of the mevalonate pathway enzymes inhibited cell growth. Gene titration studies and metabolite profiling using liquid chromatography-mass spectrometry linked the growth inhibition phenotype with the accumulation of the pathway intermediate 3-hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA). Such an accumulation implies that the activity of HMG-CoA reductase was insufficient to balance flux in the engineered pathway. By modulating HMG-CoA reductase production, we eliminated the pathway bottleneck and increased mevalonate production. These results demonstrate that balancing carbon flux through the heterologous pathway is a key determinant in optimizing isoprenoid biosynthesis in microbial hosts.

Global transcriptome response of recombinant Escherichia coli to heat-shock and dual heat-shock recombinant protein induction

Recombinant Escherichia coli cultures are used to manufacture numerous therapeutic proteins and industrial enzymes, where many of these processes use elevated temperatures to induce recombinant protein production. The heat-shock response in wild-type E. coli has been well studied. In this study, the transcriptome profiles of recombinant E. coli subjected to a heat-shock and to a dual heat-shock recombinant protein induction were examined. Most classical heat-shock protein genes were identified as regulated in both conditions. The major transcriptome differences between the recombinant and reported wild-type cultures were heavily populated by hypothetical and putative genes, which indicates recombinant cultures utilize many unique genes to respond to a heat-shock. Comparison of the dual stressed culture data with literature recombinant protein induced culture data revealed numerous differences. The dual stressed response encompassed three major response patterns: induced-like, in-between, and greater than either individual stress response. Also, there were no genes that only responded to the dual stress. The most interesting difference between the dual stressed and induced cultures was the amino acid-tRNA gene levels. The amino acid-tRNA genes were elevated for the dual cultures compared to the induced cultures. Since, tRNAs facilitate protein synthesis via translation, this observed increase in amino acid-tRNA transcriptome levels, in concert with elevated heat-shock chaperones, might account for improved productivities often observed for thermo-inducible systems. Most importantly, the response of the recombinant cultures to a heat-shock was more profound than wild-type cultures, and further, the response to recombinant protein induction was not a simple additive response of the individual stresses.