Development of Geobacillus thermoglucosidasius as a robust platform for production of chemicals from renewables through modelling and experimentation

Lead Research Organisation: Imperial College London
Department Name: Life Sciences - Biology


In this project, researchers from Imperial College London and the University of Bath will work together with the company TMO Renewables Ltd to (a) understand fundamental aspects of the physiology and biochemistry of the thermophilic bacterium Geobacillus thermoglucosidasius, which the company uses in its current bio-ethanol process, and (b) develop computer based metabolic models, using a combination of genome sequence information and experimental measurements, which will be useful for predicting how to make changes to the organism so that it can produce a specific end-product from a variety of different substrates. While the company has been successful in creating a strain of Geobacillus thermoglucosidasius that can produce ethanol from renewable lignocellulose and fermentable components of waste, this was done with little understanding of how the organism behaves under complex fermentation conditions. During this process, many observations have been made that are not easy to explain from our limited current knowledge of the organism. As well as a financial contribution to the project, the company will provide the genome sequence for their parent strain. This is the first (available) complete genome sequence for this species of thermophile and provides the academic researchers with a significant platform from which to make new discoveries. Building on this platform, the research team will apply recently-developed methods for model building, model validation and physiological investigation. The latter will involve the newly-developed approach of 'transcriptomics' by 'RNA -sequencing' to understand how the organism regulates its metabolism and behaviour under different physiological conditions. Direct analysis of RNA (strictly speaking, it has to be converted to DNA before sequencing) using modern methods of high-throughput sequencing is an advance on the previous approach using microarrays, because it does not rely on initial deduction of which are bona-fide gene sequences in a genome. Because the analysis is essentially blind to prior assumptions, it has revealed many unexpected features of regulation in different bacteria. Papers on the application of this method to bacteria only started appearing in 2009, and most of these either focus on methods development or pathogenic organisms. This project will give us the opportunity to look at an industrially relevant organism, addressing questions that are pertinent to industrial fuel and chemical production from biomass and ultimately testing hypotheses and strains in an industrial context. Therefore, there is a strong chance for discovering new and fundamental processes underlying the regulation of microbial growth and metabolism. One of the outputs from this project will be a set of metabolic models. In silico metabolic models can be useful for predicting how metabolic flux should be altered to achieve a specific outcome (eg enhanced growth or metabolite overproduction). So, as part of this exercise, we will use the models in a metabolic engineering programme to make a new metabolite, not normally produced by this strain. Using the model, we should be able to predict how flux through different pathways should be changed to accomplish the dual requirements of rapid growth and product formation. In addition to this, we hope to link the transcriptomic analysis to the models. Metabolic models are essentially static pictures that do not adequately incorporate the dynamic aspects of physiological regulation. By studying cells under different growth conditions, we can generate a set of 'condition-specific models' which can be linked through comparative analysis of the transcriptomic data. The team involves a systems biologist who is expert at integrating different types of data, who will explore the possibility of linking the two types of analysis in a meaningful manner.

Technical Summary

The need to produce fuels and chemicals from renewable lignocellulose-derived feedstocks, thereby reducing the use of fossil fuels, is generally recognised. However, there are many challenges to achieving commercial viability, ranging from cost-effective release of fermentable carbohydrates to optimised re-direction of metabolic flux. Moving beyond ethanol production, the rationale for using yeast diminishes, due to its limited substrate range. Working with the applicants, TMO Renewables have engineered the thermophile, Geobacillus thermoglucosidasius, to create a metabolically versatile ethanologen and used this as the basis for a commercially-viable process using treated municipal solid waste as a substrate. This has served not only to demonstrate the potential of the organism, but also to highlight the problems of engineering a relatively poorly characterised organism. In this project, starting with the genome sequence of G. thermoglucosidasius NCIMB 11955 provided by TMO, we will: develop and experimentally validate a genome scale metabolic model to support future metabolic engineering of this organism; explore the mechanism(s) of catabolite repression operating in both simple laboratory and complex industrial media and; combine these findings to produce a strain optimally engineered for 2-butanol production from mixed sugars in the absence of catabolite repression. While the overall aim is to increase the potential of Geobacillus spp as industrial organisms, this will be done by improving our fundamental understanding of the biochemistry and physiology of this increasingly important genus. In particular, we will apply RNA-seq, supported by gene disruption and detailed enzyme characterisation, to improve the genome annotation and explore catabolite regulation. Combined with comparative bioinformatics against Bacillus spp, we expect to rapidly highlight the differences between the genera, enabling us to focus on features specific to Geobacillus spp.

Planned Impact

Who will benefit from this research? Wider Academic Community: This work will raise our understanding of the metabolic physiology of Geobacillus spp to a level approaching that of better characterised organisms such as Bacillus subtilis. It will produce a set of freely available genomic sequences, annotated with extensive experimental support, and a similarly supported set of genome-scale metabolic models. These will be released through online databases (eg NCBI and the RAST server), publication in relevant academic journals and presentations at national and international scientific meetings. Part of this work will focus on the issue of catabolite regulation during growth in the complex mixtures of monomeric and polymeric carbohydrates found in lignocellulosic hydrolysates, which should be of interest to both academic and industrial researchers working on Cleantech processes. Commercial Private Sector: This research will directly benefit the industrial partner, TMO Renewables Ltd (TMO), which has developed an integrated process technology for converting waste biomass into valuable products. TMO's business model is to license their technology to industrial partners who own and operate the facilities. While the initial focus has been on ethanol production, TMO wish to expand the range of chemicals that can be produced from their technology platform. The ability to switch products or add extra products without a major rebuild of the original plant is very attractive and would offer a more flexible (and hence lower risk) commercial proposition to their industrial partners. More broadly, the results of this programme should benefit all companies operating in the area of chemicals from renewables. Primarily, this will be through furthering our understanding of catabolite regulation (see above). However, a number of companies (including Biocaldol and Green Biologics in the UK) recognise that thermophiles such as Geobacillus spp, that express a repertoire of glycoside hydrolases, can offer process advantages in the conversion of lignocellulosic wastes. So, we expect that both the genomic information and the metabolic engineering approach will have wide industrial interest and potential for application. By collaborating with an established industrial partner, there is a realistic opportunity to rapidly exploit the modelling, regulatory information and metabolic engineering strategies arising for the production and commercialisation of new products. TMO has recently signed a commercial contract with a US Cleantech Company (Fiberight) to build multiple community-scale plants which will convert municipal waste into ethanol. There will therefore be multiple opportunities over the next 5 years to incorporate a demonstration scale application in these commercial plants. National and International Perspective: Climate change: A primary driver for the move from fossil fuels to fuels and chemicals from waste, or sustainably derived renewables, is the reduction in greenhouse gas (GHG) emissions. An efficiently operated biorefinery using cellulosic substrates should be able to deliver an 80% reduction in GHG emissions compared to its fossil fuel equivalent (based on ethanol production). This will help meet national and international targets for use of renewables and mitigation of climate change. Green jobs: The successful delivery of this project will have an impact on delivering green jobs within the UK and further afield - a more diversified, and hence valuable, technology platform will be more attractive to new customers and take up of the technology will be greater, promoting growth within the Cleantech sector. For the PDRAs, the possibility of working closely with an established and developing Cleantech company, including spending time working at TMO, will give them an excellent perspective of both academic and industrial research environments, which should be invaluable for their future employment prospects.


10 25 50
Description Development of Geobacillus thermoglucosidasius as a robust platform for production of chemicals from renewables, through modelling and experimentation

WP1 To use the genome sequence of G. thermoglucosidasius NCIMB11955 to build and validate a genome-scale metabolic model.
The genome sequence of G. thermoglucosidasius TM242 was provided to us by TMO Renewables (note that this is the ethanol producing triple-mutant strain and not the parent strain NCIMB 11955). The original sequence was produced by Integrated Genomics and had been used for individual gene searches. A whole genome comparison with other species of Geobacillus spp showed that the assembly contained an artificial inversion, which was corrected, and the gene ordering corrected to start at the origin of replication.
It was assumed that the sequence of NCIMB11955 was the same as TM242 but without the knock-outs and knock-ins. However, discussions with Nigel Minton (Nottingham) suggest that additional mutations may have been accumulated during the development of TM242. In order to build a genome scale model for flux balance analysis it was necessary to go beyond automated annotation and to check for pathway "holes" and "dead-ends". For this we devised the tool "Pathway Booster", which allows the user to compare pathway predictions derived from whole genome sequences with up to five chosen organisms. This has revealed some unique features of Geobacillus genomics and metabolism, which will be published.
Based on a genome sequence with improved annotation, we have built a flux balance model incorporating 1011 reactions, with functions encoded by 859 genes. The directionality of 950 of these reactions was determined by thermodynamic analysis, and measured values of maintenance energy and cell composition were included. When fluxes were optimised for biomass production the predicted yields were consistent with experimental observations, validating the model.
Using the MOMA approach (minimisation of metabolic adjustment), the model was then used to predict knockouts and fluxes for production of (1) succinate and (2) 2,3 butanediol (the precursor of 2-butanol).
RNAseq information has been obtained for NCIMB11955 growing aerobically and anaerobically at 2 different growth rates and in rich vs minimal media. Obtaining good quality RNA proved to be a problem with G. thermoglucosidasius under stressed conditions. Data have been analysed and will be used to constrain the metabolic model. This remains to be completed.

WP2 To investigate carbon catabolite repression (CCR) under aerobic and anaerobic conditions.
Mutational analysis
We have carried out a site-directed mutational analysis of the catabolite repression system of G. thermoglucosidasius 11955, the wild-type strain, as this can be grown on minimal media, permitting simpler analysis of the created mutant strains (See Fig. 1).
An HPr knock-out did succeed in removing catabolite repression but the strain grew poorly on both glucose minimal media and on rich media. Deletion of the HPr kinase gene appeared to be lethal as no mutants could be obtained. The construction of an HPrS49A mutant, to prevent phosphorylation of the serine residue but to retain the histidine that is essential for glucose uptake, resulted in a wild-type phenotype that used glucose in preference to xylose. It could be that the HPr homologue, CRh, has taken over the function of HPr in the HPrS49A mutant; consequently, whilst CRhS46A exhibited wild-type behaviour, the double mutant, HPrS49A CRhS46A showed simultaneous metabolism of glucose and xylose under microaerophilic conditions in shake flasks. Moreover, the double-mutant grew better than the HPr knock-out strain and was generally more resilient and durable. It is hypothesised that the CRh did not fulfil the role of HPr in the HPr-knockout strain due to insufficient uptake of glucose, since HPr is also involved in glucose transport.

Fig 1. Proposed Geobacillus catabolite repression system based on Bacillus subtilis.

Fermentations with the double-mutant on a simulated mixture of 0.5% glucose, 0.25% xylose, 0.25% arabinose in a controlled bioreactor showed that, under aerobic conditions, the double mutant exhibited the expected phenotype of simultaneous utilisation of all three substrates (with a notable preference for arabinose over xylose) at a rate similar to that observed with glucose alone. However, under anaerobic conditions, while co-utilisation of glucose and the pentoses was observed, this appeared to be as a result of a reduced rate of glucose metabolism. This clearly demonstrates that the double mutation, HPrS49A + CRhS46A, affects anaerobic glucose metabolism in an unexpected manner.

WP3 To design and implement a metabolic engineering strategy for production of 2 butanol via 2,3-butanediol from simple carbohydrates.

Pathway construction
G. thermoglucosidasius TM242 is the genetically-engineered strain that TMO Renewables created as their ethanol-producing organism, a later strain having one of the sporulation genes disrupted. Our aim was to construct a metabolic pathway that diverted pyruvate away from ethanol production to the formation of butanol; the relevant metabolic pathways and their enzymes are shown in Fig.2.

Fig. 2. Metabolic route for biobutanol production [enzymes shown in blue type]. The existing pathway to ethanol is shown by the purple arrows.

From the metabolic model constructed for G. thermoglucosidasius (WP1 & 2), protein sequence alignments and enzyme assays, the acetolactate synthase (AS) and acetoin reductase (AR) in the organism are predicted to give RR-butanediol and not the required stereoisomer, RS butanediol, for progressing to 2-butanol. Therefore, the genes encoding an AS, an AR and an acetolactate decarboxylase (AD; this enzyme is absent from G. thermoglucosidasius) have been cloned from the thermophile Bacillus coagulans. Co expression of AS and AD in E.coli resulted in good yields of acetoin, and with the additional expression of AR, butanediol was detected with HPLC. Enzyme assays confirmed the presence of the active enzymes. The three genes have been incorporated into one plasmid (pUCG3.8) for expression in G. thermoglucosidasius; the three genes are under the control of two lactate dehydrogenase (LDH) promotors, one for AS and AD, and one for AR [one promoter for all three enzymes had shown little or no expression of AR].
Propanediol dehydratase (PD) is an enzyme complex [encoded by pduCDE] and is thought to be expressed and enclosed within a diol-dehydratase protein compartment [encoded by pduAB]. Data from Bacillus subtilis suggest that PD may well have catalytic activity with butanediol. Therefore, we have created knock-out mutants of the pduAB genes encoding the shell proteins and replaced them with the LDH promotor to upregulate pduCDE; this was performed in G. thermoglucosidasius TM236 in which the ldh and pfl genes [encoding lactate dehydrogenase and pyruvate formate lyase] have been deleted but the pyruvate dehydrogenase complex [PDHC] has not been upregulated (to reduce the flux of pyruvate to ethanol; see Fig. 2).
In a separate PhD project in the research group, a butanol dehydrogenase gene [bdh] in G. thermoglucosidasius was cloned and expressed, and the recombinant protein was shown to be enzymically active with butanone and NADH. This gene was cloned into the ldh site in G. thermoglucosidasius TM236 ?pduAB and placed under the control of the Geobacillus rpsL constitutive promoter to effect high-level production [the rpsL gene encodes the 30S ribosomal protein S12]. This strain was able to produce 2-butanol when fed with 2,3-butanediol, demonstrating that PD does have BD activity.
G. thermoglucosidasius TM236 ?pduAB with the integrated bdh has been successfully transformed with the plasmid pUCG3.8-als-ad-ar to construct the complete 2-butanol pathway; this transformant is currently being grown microaerobically, first in flasks followed by HPLC and GC analysis of the culture medium for butanol production. If successful, this will be followed by bioreactor growth under controlled redox potential, to optimise 2-butanol production. Induction of adhE required for ethanol production in TM236 requires a lower redox potential than induction of the ldh promoter, which drives expression of the butanol pathway. Therefore, even though it will ultimately be necessary to delete adhE to construct a strain which should only make 2-butanol, redox control of pathway induction should allow us to evaluate the functionality of this new pathway under true fermentative conditions. Future gene knock-out experiments to prevent the flux of pyruvate to ethanol in this transformant are planned.
This work is being prepared for submission to "Metabolic Engineering"

Plasmid constructions
The metabolic engineering to construct a butanol-producing strain has necessitated the construction of two plasmids for gene deletion and allelic exchange: pUCG3.8repB and pUCG3.8bgl.
The Geobacillus vector pUCG3.8 was made temperature sensitive by replacing the origin of replication (RepBST1) from Geobacillus stearothermophilus with that from Staphylococcus aureus (repB) to create pUCG3.8repB, which allows selection of integrative events by growth at a non-permissive temperature with kanamycin present in the medium.
pUCG3.8bgl was constructed to contain the gene encoding the Thermus thermophilus ß glucosidase gene (bgl) which, when expressed in the presence of the synthetic substrate X Glu (5-bromo-4-chloro-3-indolyl-ß-D-glucopyranoside), leads to the production of the toxic blue dye, 5,5-dibromo-4,4-dichloro-indigo. Its use in the deletion of a target gene, and the incorporation of an LDH promoter, is shown in Fig. 3, where the first crossover deletes the target gene. This event is selected on X-Glu kanamycin agar, which only permits growth of cells that have the integrated construct with kanamycin resistance and have lost the plasmid backbone encoding the Bgl enzyme. The 2nd crossover event relies on homologous recombination between the 5' flanking region of the target gene and the duplicate region incorporated from the plasmid in the 1st crossover event. This second crossover will render the cells sensitive to kanamycin and can be identified by replica plating.

Thus we have a new technique to create marker-less mutations in Geobacillus and have utilised this technique to delete the proteinaceous shell of the Geobacillus propanediol-utilization microcompartment, described above, and to incorporate the LDH promoter to control the remaining propanediol-utilizing genes. We have demonstrated that this LDH promoter is indeed functional, providing evidence that the plasmid can effect targetted allelic exchange.
This work is being prepared for publication in Appl. Environ. Microbiol.

WP4 To investigate fermentation on complex "industrial" feedstocks.
The original project envisaged technology transfer to TMO Renewables to evaluate 2-butanol production in a catabolite de-repressed strain on an industrial lignocellulosic feedstock. A number of factors have combined to limit the scope of this final objective. Firstly, TMO Renewables went into administration during the course of this project. However, in an IBTI project we have generated small quantities of hydrolysed distillers dried grain and solubles (DDGS). This was therefore used as a simulated industrial feedstock. Secondly, while we have proof of principle for construction of the 2-butanol producing strain, it requires some final tailoring and was not available in time to make the additional modifications required to produce a catabolite de-repressed strain. Hence we resolved to make a catabolite de-repressed version of the ethanol producing TM242 strain to test. However, despite the success of making the double HPrS49A CRhS46A mutant in NCIMB 11955, we were consistently unable to introduce the second mutation into TM242, regardless of which mutation was introduced first. This was tried on numerous occasions so we can confidently state that NCIMB and TM242 behave differently in this respect.
Therefore, as neither a 2-butanol nor an ethanol producing strain were available to us, we used the catabolite de-repressed version of NCIMB11955 to confirm that the results on the simulated lignocelluosic feedstock were also observed on a real feedstock.
Exploitation Route 1) By combining the 2-butanol production pathway with the mechanism to generate a catabolite repression resistant strain, it should be feasible to devise a process to produce 2-btanol from lignocellulosic feedstocks
2) The method for creating gene knockouts and allelic replacement should be generically useful
3) Metabolic models are generically useful
Sectors Manufacturing, including Industrial Biotechology

Description Catabolite repression resistant strains provided to Rebio Technologies Ltd
First Year Of Impact 2015
Sector Manufacturing, including Industrial Biotechology
Title Catabolite repression resistant strains 
Description Strains of Geobacillus thermoglucosidasius which utilise glucose and xylose together rather than sequentially 
Type Of Material Technology assay or reagent 
Year Produced 2015 
Provided To Others? Yes  
Impact Too early to say 
Title Pathway Booster 
Description Software for assigning quality on protein annotation 
Type Of Material Technology assay or reagent 
Provided To Others? No  
Impact Used by other researchers 
Description Popular Science (Bath) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Public/other audiences
Results and Impact Pint of Science - interactive 30 min presentation followed by Q &A in a local pub
Year(s) Of Engagement Activity 2016