Metabolic characterization of the chitinolytic bacterium Serratia marcescens using a genome-scale metabolic model.

BACKGROUND: Serratia marcescens is a chitinolytic bacterium that can potentially be used for consolidated bioprocessing to convert chitin to value-added chemicals. Currently, S. marcescens is poorly characterized and studies on intracellular metabolic and regulatory mechanisms would expedite development of bioprocessing applications. RESULTS: In this study, our goal was to characterize the metabolic profile of S. marcescens to provide insight for metabolic engineering applications and fundamental biological studies. Hereby, we constructed a constraint-based genome-scale metabolic model (iSR929) including 929 genes, 1185 reactions and 1164 metabolites based on genomic annotation of S. marcescens Db11. The model was tested by comparing model predictions with experimental data and analyzed to identify essential aspects of the metabolic network (e.g. 138 essential genes predicted). The model iSR929 was refined by integrating RNAseq data of S. marcescens growth on three different carbon sources (glucose, N-acetylglucosamine, and glycerol). Significant differences in TCA cycle utilization were found for growth on the different carbon substrates, For example, for growth on N-acetylglucosamine, S. marcescens exhibits high pentose phosphate pathway activity and nucleotide synthesis but low activity of the TCA cycle. CONCLUSIONS: Our results show that S. marcescens model iSR929 can provide reasonable predictions and can be constrained to fit with experimental values. Thus, our model may be used to guide strain designs for metabolic engineering to produce chemicals such as 2,3-butanediol, N-acetylneuraminic acid, and n-butanol using S. marcescens.

Engineering microbial consortia by division of labor.

During microbial applications, metabolic burdens can lead to a significant drop in cell performance. Novel synthetic biology tools or multi-step bioprocessing (e.g., fermentation followed by chemical conversions) are therefore needed to avoid compromised biochemical productivity from over-burdened cells. A possible solution to address metabolic burden is Division of Labor (DoL) via natural and synthetic microbial consortia. In particular, consolidated bioprocesses and metabolic cooperation for detoxification or cross feeding (e.g., vitamin C fermentation) have shown numerous successes in industrial level applications. However, distributing a metabolic pathway among proper hosts remains an engineering conundrum due to several challenges: complex subpopulation dynamics/interactions with a short time-window for stable production, suboptimal cultivation of microbial communities, proliferation of cheaters or low-producers, intermediate metabolite dilution, transport barriers between species, and breaks in metabolite channeling through biosynthesis pathways. To develop stable consortia, optimization of strain inoculations, nutritional divergence and crossing feeding, evolution of mutualistic growth, cell immobilization, and biosensors may potentially be used to control cell populations. Another opportunity is direct integration of non-bioprocesses (e.g., microbial electrosynthesis) to power cell metabolism and improve carbon efficiency. Additionally, metabolic modeling and 13C-metabolic flux analysis of mixed culture metabolism and cross-feeding offers a computational approach to complement experimental research for improved consortia performance.

Design and modularized optimization of one-step production of N-acetylneuraminic acid from chitin in Serratia marcescens.

Chitin is an abundant, biorenewable, nitrogen-rich biomass feedstock that can be potentially developed for biochemical production; however, efficient bioprocesses have yet to be established. Here, we demonstrate an engineered bioprocess to produce N-acetylneuraminic acid (Neu5Ac) directly from chitin using the chitinolytic bacterium, Serratia marcescens by selecting and characterizing promoters, characterization of heterologous enzyme activity, and optimization of pathway fluxes. By generating RNASeq data for S. marcescens growth in different carbon-limited conditions (glucose, N-acetylglucosamine, and glycerol), 12 promoters with varying strength were identified and characterized to implement for transcriptional control. Neu5Ac production was initially engineered into S. marcescens through heterologous expression of N-acetylglucosamine 2-epimerase (slr1975) and N-acetylneuraminic acid aldolase (nanA). The activity of both genes was characterized in vitro for kinetics and in vivo expression using promoters identified in this study. Optimization of Neu5Ac production was accomplished by balancing pathways fluxes through promoter swapping and replacing the reversible nanA with the irreversible gene neuB. The optimized recombinant strain P T5 -slr1975-P rplJ -neuB was able to produce 0.48 g/L Neu5Ac from 20 g/L N-acetylglucosamine, and 0.30 g/L Neu5Ac from 5 g/L crystal chitin. These results represent the first demonstration of direct conversion of crystal chitin to N-acetylneuraminic acid.

Study of ChiR function in Serratia marcescens and its application for improving 2,3-butanediol from crystal chitin.

Microbial utilization of chitin, a potential renewable biomass feedstock, is being pursued as a means of developing novel consolidated bioprocessing for the production of chemicals. Serratia marcescens is a gram-negative bacterium that is known for its chitinolytic capability and as a native 2,3-butanediol producer. In S. marcescens, ChiR has been suggested to be a positive regulator of chitinase production. In this study, we aim to understand the effect of ChiR in regulating nine chitinase-related genes in S. marcescens Db11 and demonstrate manipulation of chiR as a useful and efficient genetic target to enhance chitin utilization. First, a chiR overexpression (chiROE) strain and a chiR deletion (ΔchiR) strain were generated and characterized in terms of cellular growth, chitinase activity, and total secreted protein. Compared to the wild-type Db11 strain, the S. marcescens chiROE strain showed an increase in chitinase activity (2.14- to 6.31-fold increase). Increased transcriptional expression of chitinase-related genes was measured using real-time PCR, showing 2.12- to 10.93-fold increases. The S. marcescens ΔchiR strain showed decreases in chitinase activity (4.5- to 25-fold decrease), confirming ChiR's role as a positive regulator of chitinase expression. Finally, chiR overexpression was investigated as a means of increasing biochemical production (2,3-butanediol) from crystal chitin. The chiROE strain produced 1.13 ± 0.08 g/L 2,3-butanediol from 2% crystal chitin, a 2.83-fold improvement from the wild-type strain, indicating ChiR is an important and useful genetic engineering target for enhancing chitin utilization in S. marcescens.

Lack of Overt Genome Reduction in the Bryostatin-Producing Bryozoan Symbiont "Candidatus Endobugula sertula".

The uncultured bacterial symbiont "Candidatus Endobugula sertula" is known to produce cytotoxic compounds called bryostatins, which protect the larvae of its host, Bugula neritina The symbiont has never been successfully cultured, and it was thought that its genome might be significantly reduced. Here, we took a shotgun metagenomics and metatranscriptomics approach to assemble and characterize the genome of "Ca Endobugula sertula." We found that it had specific metabolic deficiencies in the biosynthesis of certain amino acids but few other signs of genome degradation, such as small size, abundant pseudogenes, and low coding density. We also identified homologs to genes associated with insect pathogenesis in other gammaproteobacteria, and these genes may be involved in host-symbiont interactions and vertical transmission. Metatranscriptomics revealed that these genes were highly expressed in a reproductive host, along with bry genes for the biosynthesis of bryostatins. We identified two new putative bry genes fragmented from the main bry operon, accounting for previously missing enzymatic functions in the pathway. We also determined that a gene previously assigned to the pathway, bryS, is not expressed in reproductive tissue, suggesting that it is not involved in the production of bryostatins. Our findings suggest that "Ca Endobugula sertula" may be able to live outside the host if its metabolic deficiencies are alleviated by medium components, which is consistent with recent findings that it may be possible for "Ca Endobugula sertula" to be transmitted horizontally. IMPORTANCE: The bryostatins are potent protein kinase C activators that have been evaluated in clinical trials for a number of indications, including cancer and Alzheimer's disease. There is, therefore, considerable interest in securing a renewable supply of these compounds, which is currently only possible through aquaculture of Bugula neritina and total chemical synthesis. However, these approaches are labor-intensive and low-yielding and thus preclude the use of bryostatins as a viable therapeutic agent. Our genome assembly and transcriptome analysis for "Ca Endobugula sertula" shed light on the metabolism of this symbiont, potentially aiding isolation and culturing efforts. Our identification of additional bry genes may also facilitate efforts to express the complete pathway heterologously.

Systematic analysis of intracellular mechanisms of propanol production in the engineered Thermobifida fusca B6 strain.

Thermobifida fusca is a moderately thermophilic actinobacterium naturally capable of utilizing lignocellulosic biomass. The B6 strain of T. fusca was previously engineered to produce 1-propanol directly on lignocellulosic biomass by expressing a bifunctional butyraldehyde/alcohol dehydrogenase (adhE2). To characterize the intracellular mechanisms related to the accumulation of 1-propanol, the engineered B6 and wild-type (WT) strains were systematically compared by analysis of the transcriptome and intracellular metabolome during exponential growth on glucose, cellobiose, and Avicel. Of the 18 known cellulases in T. fusca, 10 cellulase genes were transcriptionally expressed on all three substrates along with three hemicellulases. Transcriptomic analysis of cellodextrin and cellulose transport revealed that Tfu_0936 (multiple sugar transport system permease) was the key enzyme regulating the uptake of sugars in T. fusca. For both WT and B6 strains, it was found that growth in oxygen-limited conditions resulted in a blocked tricarboxylic acid (TCA) cycle caused by repressed expression of Tfu_1925 (aconitate hydratase). Further, the transcriptome suggested a pathway for synthesizing succinyl-CoA: oxaloacetate to malate (by malate dehydrogenase), malate to fumarate (by fumarate hydratase), and fumarate to succinate (by succinate dehydrogenase/fumarate reductase) which was ultimately converted to succinyl-CoA by succinyl-CoA synthetase. Both the transcriptome and the intracellular metabolome confirmed that 1-propanol was produced through succinyl-CoA, L-methylmalonyl-CoA, D-methylmalonyl-CoA, and propionyl-CoA in the B6 strain.

Evolutionary engineering for industrial microbiology.

Superficially, evolutionary engineering is a paradoxical field that balances competing interests. In natural settings, evolution iteratively selects and enriches subpopulations that are best adapted to a particular ecological niche using random processes such as genetic mutation. In engineering desired approaches utilize rational prospective design to address targeted problems. When considering details of evolutionary and engineering processes, more commonality can be found. Engineering relies on detailed knowledge of the problem parameters and design properties in order to predict design outcomes that would be an optimized solution. When detailed knowledge of a system is lacking, engineers often employ algorithmic search strategies to identify empirical solutions. Evolution epitomizes this iterative optimization by continuously diversifying design options from a parental design, and then selecting the progeny designs that represent satisfactory solutions. In this chapter, the technique of applying the natural principles of evolution to engineer microbes for industrial applications is discussed to highlight the challenges and principles of evolutionary engineering.

Metabolic gene-deletion strains of Escherichia coli evolve to computationally predicted growth phenotypes.

Genome-scale metabolic models have a promising ability to describe cellular phenotypes accurately. Here we show that strains of Escherichia coli carrying a deletion of a single metabolic gene increase their growth rates (by 87% on average) during adaptive evolution and that the endpoint growth rates can be predicted computationally in 39 of 50 (78%) strains tested. These results show that computational models can be used to predict the eventual effects of genetic modifications.

Laboratory evolution and multi-platform genome re-sequencing of the cellulolytic actinobacterium Thermobifida fusca.

Biological utilization of cellulose is a complex process involving the coordinated expression of different cellulases, often in a synergistic manner. One possible means of inducing an organism-level change in cellulase activity is to use laboratory adaptive evolution. In this study, evolved strains of the cellulolytic actinobacterium, Thermobifida fusca, were generated for two different scenarios: continuous exposure to cellobiose (strain muC) or alternating exposure to cellobiose and glucose (strain muS). These environmental conditions produced a phenotype specialized for growth on cellobiose (muC) and an adaptable, generalist phenotype (muS). Characterization of cellular phenotypes and whole genome re-sequencing were conducted for both the muC and muS strains. Phenotypically, the muC strain showed decreased cell yield over the course of evolution concurrent with decreased cellulase activity, increased intracellular ATP concentrations, and higher end-product secretions. The muS strain increased its cell yield for growth on glucose and exhibited a more generalist phenotype with higher cellulase activity and growth capabilities on different substrates. Whole genome re-sequencing identified 48 errors in the reference genome and 18 and 14 point mutations in the muC and muS strains, respectively. Among these mutations, the site mutation of Tfu_1867 was found to contribute the specialist phenotype and the site mutation of Tfu_0423 was found to contribute the generalist phenotype. By conducting and characterizing evolution experiments on Thermobifida fusca, we were able to show that evolutionary changes balance ATP energetic considerations with cellulase activity. Increased cellulase activity is achieved in stress environments (switching carbon sources), otherwise cellulase activity is minimized to conserve ATP.

Identification of metabolic changes in genetically unstable stem cells by using model analysis of gene expression.

Stem-cell research seeks to address many different questions related to fundamental stem-cell function with the ultimate goal of being able to control and utilize stem cells for a broad range of therapeutic needs. While a large amount of work is focused on discovering and controlling differentiation mechanisms in stem cells, an equally interesting and important area of work is to understand the basics of stem-cell propagation and self-renewal. With high-throughput genomics and transcriptomic information on hand, it is becoming possible to address some of the detailed mechanistic processes occurring in stem cells, though interpretation of these data is often difficult. In this work, stem cells with genetic abnormalities were compared to genetically normal stem cells using gene-expression array data integrated with a large-scale metabolic model to help interpret changes in metabolism resulting in the identification of several metabolic pathways that were different in the normal and abnormal cells.

Metabolic engineering of Thermobifida fusca for direct aerobic bioconversion of untreated lignocellulosic biomass to 1-propanol.

Biofuel production from renewable resources can potentially address lots of social, economic and environmental issues but an efficient production method has yet to be established. Combinations of different starting materials, organisms and target fuels have been explored with the conversion of cellulose to higher alcohols (1-propanol, 1-butanol) being one potential target. In this study we demonstrate the direct conversion of untreated plant biomass to 1-propanol in aerobic growth conditions using an engineered strain of the actinobacterium, Thermobifida fusca. Based upon computational predictions, a bifunctional butyraldehyde/alcohol dehydrogenase was added to T. fusca leading to 1-propanol production during growth on glucose, cellobiose, cellulose, switchgrass and corn stover. The highest 1-propanol titer (0.48g/L) was achieved for growth on switchgrass. These results represent the first demonstration of direct conversion of untreated lignocellulosic biomass to 1-propanol in an aerobic organism and illustrate the potential utility of T. fusca as an aerobic, cellulolytic bioprocess organism.

Development and application of a PCR-targeted gene disruption method for studying CelR function in Thermobifida fusca.

Thermobifida fusca is a high-G+C-content, thermophilic, Gram-positive soil actinobacterium with high cellulolytic activity. In T. fusca, CelR is thought to act as the primary regulator of cellulase gene expression by binding to a 14-bp inverted repeat [5'-(T)GGGAGCGCTCCC(A)] that is upstream of many known cellulase genes. Previously, the ability to study the roles and regulation of cellulase genes in T. fusca has been limited largely by a lack of established genetic engineering methods for T. fusca. In this study, we developed an efficient procedure for creating precise chromosomal gene disruptions and demonstrated this procedure by generating a celR deletion strain. The celR deletion strain was then characterized using measurements for growth behavior, cellulase activity, and gene expression. The celR deletion strain of T. fusca exhibited a severely crippled growth phenotype with a prolonged lag phase and decreased cell yields for growth on both glucose and cellobiose. While the maximum endoglucanase activity and cellulase activity were not significantly changed, the endoglucanase activity and cellulase activity per cell were highly elevated. Measurements of mRNA transcript levels in both the celR deletion strain and the wild-type strain indicated that the CelR protein potentially acts as a repressor for some genes and as an activator for other genes. Overall, we established and demonstrated a method for manipulating chromosomal DNA in T. fusca that can be used to study the cellulolytic capabilities of this organism. Components of this method may be useful in developing genetic engineering methods for other currently intractable organisms.

Toward engineering synthetic microbial metabolism.

The generation of well-characterized parts and the formulation of biological design principles in synthetic biology are laying the foundation for more complex and advanced microbial metabolic engineering. Improvements in de novo DNA synthesis and codon-optimization alone are already contributing to the manufacturing of pathway enzymes with improved or novel function. Further development of analytical and computer-aided design tools should accelerate the forward engineering of precisely regulated synthetic pathways by providing a standard framework for the predictable design of biological systems from well-characterized parts. In this review we discuss the current state of synthetic biology within a four-stage framework (design, modeling, synthesis, analysis) and highlight areas requiring further advancement to facilitate true engineering of synthetic microbial metabolism.

Influence of culture aeration on the cellulase activity of Thermobifida fusca.

Currently, one of the hurdles hindering efficient production of cellulosic biofuel is the recalcitrant nature of cellulose to hydrolysis. A wide variety of cellulase enzymes are found natively in microorganisms that can potentially be used to effectively hydrolyze cellulose to fermentable sugars. In this study, phenomenological and mechanistic parameters affecting cellulase activity were studied using the moderately thermophilic, aerobic, and cellulolytic microorganism Thermobifida fusca. Two major sets of experiments were conducted to (1) study the mechanistic differences in growth in a flask compared to a bioreactor and (2) study the cell culture parameters influencing cellulase activity using a series of bioreactor experiments. Specific cellulase and specific endoglucanase activities were found to be higher in the bioreactor as compared to flask growth. Measurements of messenger RNA transcript levels of 18 cellulase-related genes and intracellular ATP levels indicated that measured enzyme activity was likely more influenced by post-transcriptional energetics rather than transcriptional regulation. By delineating the effects of culture aeration and stir speed using a bioreactor, it was found that cellulase activity increased with increasing aeration and increasing stir speeds (highest K(l)a) with a tradeoff of decreased cellular growth at the highest stir speeds tested (400 rpm). Overall, these results allude to a connection between aeration and oxidative respiration that lead to increased ATP allowing for increased cellulase synthesis as the primary constraint on overall cellulase activity.

Exploring biodiversity for cellulosic biofuel production.

Industrial production of solvents such as EtOH and BuOH from cellulosic biomass has the potential to provide a sustainable energy source that is relatively cheap, abundant, and environmentally sound, but currently production costs are driven up by expensive enzymes, which are necessary to degrade cellulose into fermentable sugars. These costs could be significantly reduced if a microorganism could be engineered to efficiently and quickly convert cellulosic biomass directly to product in a one-step process. There is a large amount of biodiversity in the number of existing microorganisms that naturally possess the enzymes necessary to convert cellulose to usable sugars, and many of these microorganisms can directly ferment sugars to EtOH or other solvents. Currently, the vast majority of cellulolytic organisms are poorly understood and have complex metabolic networks. In this review, we survey the current state of knowledge on different cellulases and metabolic capabilities found in various cellulolytic microorganisms. We also propose that the use of large-scale metabolic models (and associated analyses) is potentially an ideal means for improving our understanding of basic metabolic network function and directing metabolic engineering efforts for cellulolytic microorganisms.

A genome-scale metabolic model of Cryptosporidium hominis.

The apicomplexan Cryptosporidium is a protozoan parasite of humans and other mammals. Cryptosporidium species cause acute gastroenteritis and diarrheal disease in healthy humans and animals, and cause life-threatening infection in immunocompromised individuals such as people with AIDS. The parasite has a one-host life cycle and commonly invades intestinal epithelial cells. The current genome annotation of C. hominis, the most serious human pathogen, predicts 3884 genes of which ca. 1581 have predicted functional annotations. Using a combination of bioinformatics analysis, biochemical evidence, and high-throughput data, we have constructed a genome-scale metabolic model of C. hominis. The model is comprised of 213 gene-associated enzymes involved in 540 reactions among the major metabolic pathways and provides a link between the genotype and the phenotype of the organism, making it possible to study and predict behavior based upon genome content. This model was also used to analyze the two life stages of the parasite by integrating the stage-specific proteomic data for oocyst and sporozoite stages. Overall, this model provides a computational framework to systematically study and analyze various functional behaviors of C. hominis with respect to its life cycle and pathogenicity.

Leveraging genome-scale metabolic models for human health applications.

Genome-scale metabolic modeling is a scalable and extensible computational method for analyzing and predicting biological function. With the ongoing improvements in computational methods and experimental capabilities, genome-scale metabolic models (GEMs) are demonstrating utility in addressing human health applications. The initial areas of highest impact are likely to be health applications where disease states involve metabolic changes. In this review, we focus on recent application of GEMs to studying cancer and the human microbiome by describing the enabling methodologies and outcomes of these studies. We conclude with proposing some areas of research that are likely to arise as a result of recent methodological advances.

Computational Modeling of the Human Microbiome.

The impact of microorganisms on human health has long been acknowledged and studied, but recent advances in research methodologies have enabled a new systems-level perspective on the collections of microorganisms associated with humans, the human microbiome. Large-scale collaborative efforts such as the NIH Human Microbiome Project have sought to kick-start research on the human microbiome by providing foundational information on microbial composition based upon specific sites across the human body. Here, we focus on the four main anatomical sites of the human microbiome: gut, oral, skin, and vaginal, and provide information on site-specific background, experimental data, and computational modeling. Each of the site-specific microbiomes has unique organisms and phenomena associated with them; there are also high-level commonalities. By providing an overview of different human microbiome sites, we hope to provide a perspective where detailed, site-specific research is needed to understand causal phenomena that impact human health, but there is equally a need for more generalized methodology improvements that would benefit all human microbiome research.

Recent advances on constraint-based models by integrating machine learning.

Research that meaningfully integrates constraint-based modeling with machine learning is at its infancy but holds much promise. Here, we consider where machine learning has been implemented within the constraint-based modeling reconstruction framework and highlight the need to develop approaches that can identify meaningful features from large-scale data and connect them to biological mechanisms to establish causality to connect genotype to phenotype. We motivate the construction of iterative integrative schemes where machine learning can fine-tune the input constraints in a constraint-based model or contrarily, constraint-based model simulation results are analyzed by machine learning and reconciled with experimental data. This can iteratively refine a constraint-based model until there is consistency between experimental data, machine learning results, and constraint-based model simulations.