Background: Actinobacillus succinogenes is a promising bacterial catalyst for the bioproduction of succinic acid from low-cost raw materials. In this work, a genome-scale metabolic model was reconstructed and used to assess the metabolic capabilities of this microorganism under producing conditions. Results: The model, iBP722, was reconstructed based on the functional reannotation of the complete genome sequence of A. succinogenes 130Z and manual inspection of metabolic pathways, covering 1072 enzymatic reactions associated with 722 metabolic genes that involve 713 metabolites. The highly curated model was effective in capturing the growth of A. succinogenes on various carbon sources, as well as the SA production under various growth conditions with fair agreement between experimental and predicted data. Calculated flux distributions under different conditions show that a number of metabolic pathways are affected by the activity of some metabolic enzymes at key nodes in metabolism, including the transport mechanism of carbon sources and the ability to fix carbon dioxide. Conclusions: The established genome-scale metabolic model can be used for model-driven strain design and medium alteration to improve succinic acid yields. Keywords: Genome-scale metabolic reconstruction, Constraints-based flux analysis, Succinic acid fermentation Background amounts of other organic acids were also produced, in- Actinobacillus succinogenes is a gram-negative facultative creasing the downstream processing costs, which makes anaerobic bacterium and is one of the major natural pro- this bioprocess less competitive. Currently, SA is mainly ducers of succinic acid (SA). It can grow on a broad produced from petrochemical feedstocks through the range of substrates, including arabinose, cellobiose, fruc- hydrogenation of maleic acid or maleic anhydride . tose, galactose, glucose, lactose, maltose, mannitol, man- However, the bio-based production using low pH yeast nose, sucrose and xylose, producing a mixture of fermentation [17–19] or anaerobic fermentation using by-products (e.g., SA, formic acid (FA), acetic acid (AA), bacteria [20–23] has been successfully implemented by and ethanol (EtOH)) as the main by-products [1, 2]. Its companies like Myriant , BASF  or BioAmber − 1 tolerance to high sugar concentrations (up to 160 g.L , offering economically and ecologically attractive al- of glucose ) and high levels of organic acids , as ternatives to the conventional petro-based SA produc- well as its capnophilic nature , make this microorgan- tion [27–29]. Some examples of SA producing systems ism potentially interesting for the production of SA at are given in Table 1, including naturally-producing the industrial scale. High-titer succinate production bacteria like Basfia succiniciproducens and Mannheimia using low-cost feedstocks like cane molasses or corn succiniproducens and genetically engineered organisms straw has been obtained [2, 6–15]; however, significant such as E. coli or S. cerevisiae.Sofar,natural producers appear to outperform most engineered strains, but devel- opments in strain design and fermentative processes are * Correspondence: firstname.lastname@example.org expected to promote the production of bio-based SA by SilicoLife Lda, Rua do Canastreiro 15, 4715-387 Braga, Portugal metabolically engineered microorganisms. For instance, Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Pereira et al. BMC Systems Biology (2018) 12:61 Page 2 of 12 Table 1 Some examples of SA bio-production systems using natural producers or metabolic engineered organisms Organism Genetic modifications Culture conditions Carbon SA Yield References −1 sources (CS) (g.g ) CS B. succiniciproducens Anaerobic, continuous Glycerol 1.02  M. succiniproducens Anaerobic, batch Glucose 0.59  M. succiniproducens – Deletion of ldhA, pflB, pta, and ackA genes Anaerobic, fed-batch Glucose 0.76  A. succinogenes Anaerobic, continuous Xylose 0.80  A. succinogenes Anaerobic, batch Glucose 0.74  A. succinogenes Anaerobic, repeated fed-batch Glucose 0.88  E. coli – Deletion of pflB, ldhA, ppc genes Anaerobic, batch Corn stalk hydrolysate 1.02  – Heterologous expression of pckA from B. subtilis E. coli – Deletion of iclR, icd, sdhAB, ackA-pta, poxB Aerobic, batch Glucose 0.72  – Heterologous expression of pepc gene from Sorghum vulgare S. cerevisiae – Deletion of sdh3, ser3, ser33 Aerobic, batch Glucose 0.05  – Overexpression of ICL1 S. cerevisiae – Deletion of pdc1, pdc5, pdc6, fum1, gpd1 Aerobic, batch Glucose 0.14  – Overexpression of pyc2, mdh3, fumC and frds1 − 1 Mass yields are given in g.g of carbon source (CS) M. succiniproducens has been metabolically engineered by production and several studies have used both S. ce- removing competing pathways, resulting in an increase in revisiae and E. coli to exploit these metabolic pathways the SA yield from 0.45 to 0.76 g of SA per gofglucose . [38, 40]. The redirection of the carbon flux through the Under optimized conditions, the A. succinogenes glyoxylate shunt provides some advantages over the oxi- − 1 wild-type strain is able to produce up to 98 g.L of SA dative TCA cycle, mainly because the decarboxylation of with an approximate yield of 90% (w/w) on glucose . isocitrate to succinyl-CoA leads to carbon loss  and, The optimization of bioprocesses has proven to further in the case of S. cerevisiae, the SA channelling from the increase SA production by using high concentrations of mitochondria to cytosol is avoided as TCA cycle en- carbon dioxide (CO ) and/or hydrogen (H )[32, 33]. zymes are located in mitochondria. However, if a reduc- 2 2 Other studies have shown that the redox state of the fer- tive TCA pathway is used, a 2-fold maximum theoretical − 1 mentation broth affects SA production, which can be yield (2 mol.mol of glucose) can be achieved com- − 1 improved by manipulating the supplementation of oxi- pared to the oxidative route (1 mol.mol of glucose) dant and reducing agents [34, 35]. Yet, due to the accu- . Many organisms, such as E. coli and S. cerevisiae mulation of other fermentative by-products, SA yields have been tested for SA production under anaerobic are still far below the maximum theoretical yield of conditions using the reductive branch of the TCA cycle − 1 1.12 g.g of glucose consumed (Y ). A com- [38, 42]. However, reducing power limitations (i.e. SA/Glc prehensive understanding of the metabolism and the NADH) or redox balance issues have shown to have phenotypic responses to environmental perturbations is an impact on the final SA yields . Metabolic en- a major step for developing efficient bioprocesses for SA gineering strategies driven by in silico modelling may production. allow to overcome these disadvantages, both under Genome-scale metabolic models (GSMMs) have aerobic and anaerobic conditions. Strategies for in- proven to be powerful tools for understanding and creasing energy and/or cofactor pools [43–45], to re-designing the metabolism of microbial strains. For in- overcome enzyme limitations  or to decrease stance, the optimization of SA production in E. coli or S. by-product generation  are just a few examples cerevisiae has been achieved by applying metabolic en- that have been exploited . gineering strategies supported by in silico modelling of A. succinogenes produces SA anaerobically through the metabolic networks [37–40]. There are three main reductive branch of the TCA cycle (i.e., C4 pathway) pathways for SA biosynthesis, including the tricarboxylic using fumarate as the final electron acceptor, which acid (TCA) cycle in the oxidative direction, the glyoxy- makes this metabolic branch highly dependent on the late shunt and the reductive TCA pathway . Typic- redox state of cultures. Phosphoenolpyruvate (PEP) node ally, under aerobic conditions, either the oxidative TCA controls the amount of flux that is directed towards the cycle or the glyoxylate shunt can be used for SA C4 and C3 pathways, adjusting the level of fermentative Pereira et al. BMC Systems Biology (2018) 12:61 Page 3 of 12 products generated by each pathway. Other metabolic obtained with the respective information, including nodes like oxaloacetate (OAA) and malate (MAL) have scores and protein features that may contain shown to link the C4 and C3 pathways via decarboxylat- Enzyme Commission (EC) numbers. ing enzymes, i.e. oxaloacetate decarboxylase and the (2) The computation of “functional scores” based on NADPH-dependent malic enzyme, respectively. The split BLAST and HMMER scores. As the range of of carbon flux at these branching points is largely influ- enzymatic functions attributed to each CDS can enced by various factors, such as the availability of CO , vary from tool to tool, a second score was pH or carbon sources. It has been previously shown that computed, i.e. so-called functional score, in order to an increase in the concentration of dissolved CO in the propose the best candidate metabolic functions. Be- fermentation broth through the supplementation of sides BLAST and HMMER scores, a functional magnesium carbonate  or sodium bicarbonate  score was computed similarly to the method used can promote an increase in the carbon flow toward the by Merlin , which is based on the frequency of C4 pathway, thus increasing the production of SA. The the EC number in the homology hits and on the presence of carbonic anhydrases (putative coding taxonomy distance between the target strain and gene, Asuc_1199), which interconverts CO and bicar- other strains within hit results. This weighed func- bonate (HCO ), may also contribute to increase CO tional scores range between 0 and 1 (1 correspond- 3 2 fixation [12, 49]. ing to a high confidence score). Here, a GSMM (named iBP722) representing a wide (3) Assignment of putative metabolic function(s). EC range of metabolic capabilities of A. succinogenes 130Z is numbers associated with the highest functional proposed. The model allows predicting and analysing the score are automatically attributed to each CDS, as impact of stoichiometric and physiological constraints well as the corresponding metabolic reaction(s) known to apply at steady-state conditions. Although the from an internal reactions database. central carbon metabolism of A. succinogenes 130Z has (4) The assignment of metabolic functions to each CDS been comprehensively described [32, 33, 50, 51], the over- is manually revised, as automatic assignments may all representation of metabolic pathways associated with fail when more than one high scoring EC number is various catabolic and anabolic capabilities of the organism found and/or EC number(s) are incorrectly are now made available. The biosynthetic pathways for vi- associated in databases. Therefore the user is tamins, cofactors and other biomass building blocks are allowed to inspect all putative assignments and described, as well as respiratory and energy consuming as- select the most appropriate or modify the similatory pathway. This model provides a detailed insight functional assignment, which can be based on on the metabolism of A. succinogenes that can be system- previous knowledge or other assumptions defined atically explored to improve the bioproduction of SA. by the user. Furthermore, more than one CDS can be associated to the same EC number, which in Methods many cases consists of subunits of the same Metabolic functional annotation multimeric enzyme. The classification of multimeric The complete genome sequence of A. succinogenes 130Z or monomeric subunits is also defined at this stage (GenBank accession number NC_009655)  was used using an internal database. for the functional annotation of genes based on hom- (5) After validation of functional assignments and ology searching methods. The annotated genes with po- enzyme subunits, the association of metabolic tential metabolic roles were manually inspected and reactions is carried out based on the associated EC associated with the corresponding coding enzyme(s) and number(s) and/or previous knowledge using as a biochemical reaction(s). An internally developed plat- reference an internal reactions database. This step is form was used to compute, assign and curate gene meta- perhaps the most critical during the reconstruction bolic functions. This platform couples automated of the metabolic network, as it will define the set of annotation tools with manual curation procedures that stoichiometric reactions that characterize a specific allows the assignment of metabolic functions to coding organism. sequences (CDSs) of a particular genome. The pipeline consists in five main steps (see Fig. 1): To note that this annotation pipeline is flexible enough, such that unassigned CDSs in step (3) can be later (1) The application of homology search tools like reviewed, particularly during the gap filling process. BLAST  and HMMER  against sequence databases, such as UniProt  to find the best Construction of the metabolic network alignment between sequences. A ranked list of hits The GSMM was initially constructed by compiling the an- with the most significant matches to each query is notated metabolic genes and their corresponding coding Pereira et al. BMC Systems Biology (2018) 12:61 Page 4 of 12 Fig. 1 Metabolic functional annotation of CDSs. The implemented framework uses an automated annotation tool coupled with manual curation procedures that allows the assignment of metabolic functions to CDSs of a particular genome. The pipeline comprises five main steps: (1) homology search CDSs against databases like UniProt; (2) computation of functional scores based on similarity; (3) assignment of putative metabolic function(s); (4) manual curation of functions; and finally (5) association of metabolic reactions enzyme(s) and biochemical reaction(s). Additionally, network. Then, each dead-end metabolite was spontaneous and transport reactions from databases like inspected to search for metabolic reactions that KEGG  and Transporter Classification Database consume, produce or transport this metabolite. (TCDB)  were added. Some metabolic reactions, al- Typically, MetaCyc or KEGG databases were though not associated with genes, were also included due used for gap-filling, i.e. to identify the sets of to evidences found in literature. The preliminary draft biochemical reactions that link each dead-end to model was then processed to: a metabolite in the network. When several alter- natives are found, a manual inspection is re- Identify metabolic gaps (or missing reactions) quired and sequence-based homology searches that either consume or produce isolated (or using one or more amino acid sequences col- dead-end) metabolites within the metabolic lected from potential candidates are used to find Pereira et al. BMC Systems Biology (2018) 12:61 Page 5 of 12 the most likely reactions set in the metabolic gene-reaction relationships) were compiled to build a draft network of A. succinogenes 130Z. metabolic model, (3) which was thereafter completed and Infer and correct the mass and charge balance of corrected by defining a biomass reaction, identifying net- biochemical reactions. Stoichiometric coefficients of work gaps and correcting inconsistencies when comparing compounds in reactions are corrected, such that the with reported information. reaction is balanced for mass and charge, usually by The final iBP722 model consists of 722 unique genes adding missing protons or water molecules. (open reading frame (ORF) coverage − 35%), 1072 reac- Identify and correct the reversibility of reactions tions and 713 unique metabolites. The model is available based on their thermodynamic properties and as a Systems Biology Markup Language (SBML) file at information found in literature. Databases like http://darwin.di.uminho.pt/models and BioModels data- MetaCyc  and tools like eQuilibrator  base  assigned with the identifier MODEL1804130001 were used. and detailed information on the curated metabolic network Include a biomass reaction representing the basic can be found in Additional file 2. macromolecular composition of A. succinogenes in terms of proteins, DNA, RNA, lipopolysaccharide Model validation (LPS), phospholipids, peptidoglycan, glycogen and Predicted growth on different carbon sources cofactors and vitamins (CAV). The synthesis of each The iBP722 model was inspected for the ability to simu- macromolecule was also represented by individual late the A. succinogenes growth on different conditions. reactions considering the building blocks molar Model simulations were performed using the OptFlux composition. For instance, the synthesis of one gram software  applying FBA-based methods that maximize of protein was calculated based on the average the biomass reaction under defined conditions, i.e. ex- amino acids composition using the set of encoded change fluxes were constrained to specific values (usually proteins of the A. succinogenes 130Z genome, experimentally measured fluxes) that allow testing growth according to the methodology proposed in . The under defined environmental conditions, such as sole car- synthesis of other cellular components, like CAV, bon sources. Some exchange fluxes like those associated + + was calculated assuming that each small molecule is with the exchange of CO ,NH ,Pi,H , vitamins and 2 4 equally present in one gram of CAV. Detailed trace elements were kept unconstrained (i.e. unlimited up- information on the biomass composition can be take rates) to provide unlimited basic nutrients for bio- found in Additional file 1. mass synthesis. A. succinogenes is auxotrophic for three amino acids: Constraints-based flux analysis L-glutamate, L-cysteine and L-methionine. Glutamate Basic stoichiometric modelling methods, such as parsi- auxotrophy is due to the organism’s inability to monious flux balance analysis (pFBA) and flux variability synthesize α-ketoglutarate, since the genes encoding for analysis (FVA), were used to interrogate the metabolic isocitrate dehydrogenase and α-ketoglutarate oxidore- properties and capabilities of the reconstructed GSMM ductase enzymes are absent in the genome. A. succino- for A. succinogenes 130Z under varying environmental genes possesses most of the genes encoding enzymes conditions. Phenotype simulations were performed by associated with the cysteine biosynthetic pathway, but maximizing the biomass reaction assuming growth under the absence of an adenylsulfate kinase to assimilate sul- defined conditions, i.e. defined minimum media contain- fate prevents the synthesis of hydrogen sulfide required ing basic components required for biomass synthesis, such for the synthesis of L-cysteine. This organism also lacks as vitamins, minerals and trace metals and explicit carbon, several genes for the biosynthesis of methionine as iden- nitrogen and sulphur sources. Maximum theoretical prod- tified during the functional annotation process. Given uct yields were calculated by maximizing the target prod- that, exchange fluxes associated with amino acid auxot- uct instead, ignoring the formation of biomass and ATP rophies in A. succinogenes, such as L-cysteine and maintenance requirements, such that the costs of product L-methionine, were maintained unconstrained, except biosynthesis in terms of carbon, energy and reducing for L-glutamate that was limited to a minimum flux equivalents were properly evaluated. value to support growth requirements and avoid glutam- ate consumption as an additional carbon source. Further Results flux constraints were introduced when inspecting carbon A. succinogenes genome-scale metabolic network flux distributions in the central carbon metabolism, and The construction of the GSMM for A. succinogenes 130Z are further detailed in Table SI 11 (Additional file 2). For was carried out in three different phases: (1) first, meta- instance, succinyl-CoA ligase reaction was limited to a bolic functions were assigned to genes; (2) then, biochem- zero flux to avoid the formation of succinic acid from ical reactions and enzymatic complexes (assigning proper succinyl-CoA, in order to be consistent with in vivo Pereira et al. BMC Systems Biology (2018) 12:61 Page 6 of 12 observations . Assuming these flux constraints, C3 and C4 metabolic pathways in A. succinogenes was growth predictions were computed and compared with examined. Given a set of flux constraints (e.g. substrate in vivo observations [4, 27]. uptake rates, q ), the predicted flux spans for the main Growth phenotypes on 22 different carbon sources reactions in the central carbon metabolism were calcu- under anaerobic conditions were tested and compared lated (Fig. 3a). Flux spans are given by the difference be- with in vivo growth data from  (Table 2 and tween the maximum and minimum predicted flux values Additional file 3). of each metabolic reaction while maintaining 95% of the The model predicted accurately growth on more than maximum biomass formation. As shown in Fig. 3b, the 90% of the carbon sources, with only two not supporting largest flux spans were associated with reactions around in silico growth (FN = 2). In silico growth on the PEP node (e.g. PEP carboxykinase (PPCK) and pyru- beta-gentiobiose and D-arabitol was not predicted, as vate kinase (PYK)), while reactions associated to the catabolic and transport reactions were not identified in Embden-Meyerhoff-Parnas pathway (glucose-6-pho- A. succinogenes. Further information on transport activ- sphate isomerase (PGI), phosphofructokinase (PFK) and ities included in the model is given in Additional file 4. enolase (ENO)) presented the lowest flux spans. Meta- bolic flux data estimated from C-labeling experiments Predicted yields for fermentation products  was further used to assess the accuracy of predicted The iBP722 model was further validated by predicting flux spans. Most in vivo measurements were between es- anaerobic production yields and comparing with experi- timated flux ranges, except for the CO exchange and mental data from batch and/or chemostat cultures of A. formate dehydrogenase (FDHmq) reactions, indicating succinogenes 130Z growing on glucose or xylose at dif- that the NADH-producing FDHmq reaction should be ferent initial concentrations. [32, 51, 63] (Fig. 2). For active and CO exchange flux should be lower. Changes each condition, carbon uptake rates were defined based in flux constraints associated with CO uptake (from un- − 1 − 1 on experimental values, except for condition C that was limited to a maximum of 4 mmol.gDCW .h ) altered − 1 − 1 set to 8 mmol.gDCW .h , and then predicted produc- predicted flux spans (Fig. 3c), especially for FDHmq, tion rates were used to calculate minimum and max- pyruvate-formate lyase (PFL) and the FA exchange reac- imum FVA yields for biomass, SA, AA, FA and EtOH tion (EX_FA), indicating a higher flexibility in metabolic (Yx/S, YSA/S, YAA/S, YFA/S, YEtOH/S, respectively). activities linked to FA accumulation. FVA spans are given by the difference between the max- The production of SA in A. succinogenes is influenced imum and minimum predicted yields while maintaining by several factors, namely the utilized carbon sources 95% of the maximum biomass formation.  or the availability of CO [5, 33, 48]. The iBP722 FVA predictions indicate higher yields for C4 model was investigated for predicting the metabolic by-products (i.e. SA) compared to C3 by-products (i.e. flexibility associated with the production of SA when AA, EtOH and FA), which is in good agreement with ex- changing CO availability or carbon sources, as well as perimental data; however predicted mass ratios between the reversibility of metabolic reactions like malic enzyme fermentative by-products, specifically SA/AA and SA/FA, (ME2) (Fig. 4). FVA yields for maximum growth simula- are higher compared with in vivo observations, especially tions show that the production of reduced by-products, when considering maximum FVA ratios. Yet, FVA spans particularly EtOH, changes with the carbon source. As indicate a significant flexibility for these ratios, which presented in Fig. 4a, the minimum and maximum FVA might explain variations in the in vivo observations. yields for ALCD2x under D-sorbitol growth conditions − 1 (0.62 and 0.69 mol.mol ) were higher compared to glu- − 1 Improving model predictions cose (0.14 and 0.22 mol.mol , respectively), with the An FVA analysis was performed to elucidate these dis- consequent accumulation of higher amounts of EtOH. crepancies and the metabolic flexibility associated with Under glucose conditions most of the carbon flux through the C3 branch is redirected toward the produc- tion of AA instead, via acetate kinase (ACKr) with the Table 2 Comparison of growth predictions and in vivo tests on production of ATP (minimum and maximum FVA yields 22 carbon sources − 1 of 0.47 and 0.55 mol.mol , respectively). Interestingly, In vivo however, is that predicted FVA yields for SA production Growth No growth hardly change between glucose and D-sorbitol growth In silico Growth TP = 19 (86%) FP = 0 (0%) conditions when considering the same transport mech- No growth FN = 2 (9%) TN = 1 (5%) anism, i.e. PEP:sugar phosphotransferase system True Positives (TP) and True Negatives (TN) indicate the number of carbon (PEP:PTS). Yet, assuming that both transport mecha- sources in which growth phenotypes were correctly predicted, while False nisms could be active (i.e., symport and PEP:PTS) for Positives (FP) and False Negatives (FN) indicate the number carbon sources in which in silico predictions did not match in vivo observations glucose uptake would increase SA minimum and Pereira et al. BMC Systems Biology (2018) 12:61 Page 7 of 12 (A1) (A2) (B1) (B2) (C1) (C2) Fig. 2 Experimental values (filled dots) versus predicted FVA yields (floating bars) from A. succinogenes 130Z cultures on glucose (A and B) and xylose (C). Predicted minimum and maximum FVA yields for biomass, SA, AA, FA and EtOH (Y Y Y ,Y ,Y , respectively) were x/S, SA/S, AA/S FA/S EtOH/S estimated while maintaining 95% of the maximum biomass formation (A1, B1 and C1). Similarly, minimum and maximum FVA yields between fermentative by-products (SA/AA, FA/AA and SA/FA) were estimated while maintaining 95% of the maximum biomass formation (A2, B2 and C2). In silico predictions were performed by setting the substrate uptake rate (q ) to the experimental value (except in condition C that was set to − 1 − 1 8 mmol.gDCW .h ). Experimental parameters for conditions A and B were obtained from batch cultures with defined medium (AM3) supplemented with 50 mM glucose and 150 mM NaHCO under anaerobic conditions [32, 51]; while parameters for C were obtained from a − 1 − 1 − 1 continuous culture at a dilution rate of 0.05 h under anaerobic conditions with supplemented medium (6 g.L yeast extract, 10 g.L corn − 1 − 1 steep liquor and 50–85 g.L xylose). No biomass yield was experimentally determined for this condition . Mass yields are given in g.g maximum FVA yields from 0.63 and 0.67 to 0.82 and FVA results (Fig. 4b) show that SA production should − 1 0.96 mol.mol , respectively. increase when ME2 occurs in the reverse direction, es- The flux exchange between the C3 and C4 branches pecially for D-sorbitol conditions, redirecting most of has been also investigated as a major factor affecting the the PYR pool toward the C4 branch. Consequently, a metabolic flexibility of sugars fermentation in A. succino- greater metabolic flexibility in SA production, especially genes . The decarboxylation of L-malate to pyruvate under glucose growth conditions, was predicted. (PYR) reducing NADP to NADPH by ME2, may have a The availability of CO was also shown to influence major role in the fermentative metabolism. Although the the production of SA (Fig. 4c). Model predictions indi- thermodynamics of this reaction is not conclusive re- cated that maximum SA yields on glucose can decrease garding its reversibility, the metabolic network was nearly 20% when decreasing the maximum CO uptake tested using both the forward and the reverse directions. rates by 50%. A shift in carbon flux distributions is Pereira et al. BMC Systems Biology (2018) 12:61 Page 8 of 12 ab Fig. 3 Metabolic flux spans of the A. succinogenes central carbon metabolism. The FVA analysis covered individual reactions represented in (a). Flux spans (represented by floating coloured bars) define the flux range of individual reactions while maintaining 95% of the maximum biomass − 1 − 1 13 formation (b) and further constraining the maximum uptake rate of CO to 4 mmol.gDCW .h (c). In vivo flux measurements from C-labeling experiments  (represented by filled dots with error bars) are also depicted. observed when decreasing CO uptake levels, redirecting enter the cell mainly through active transport systems most of the carbon flux toward the production of C3 (Additional file 4). The only two carbon sources with in- by-products like FA and EtOH. correct growth predictions were β-gentiobiose and D-arabitol, due to the absence of transport and catabolic Discussion pathways in the model, since no coding proteins were GSMMs are powerful tools to explore the metabolism of found. On the other hand, no-growth predictions for biological systems. In this work, the iBP722 model of A. glycerol were correctly identified, though transport and succinogenes 130Z was reconstructed and used as a plat- catabolic reactions for glycerol consumption are present form for the in silico analysis of the metabolic behaviour in the model. According to in vivo growth experiments of this organism during anaerobic growth. The major , no cellular growth on glycerol as a sole carbon end-product is SA, but significant amounts of other source is observed under anaerobic conditions, possibly by-products such as EtOH, FA and AA are also accumu- caused by redox imbalance under anaerobiosis. However, lated. The possibility to predict and adjust the fermenta- the addition of external electron acceptors like dimethyl- tive metabolism of A. succinogenes 130Z under different sulfoxide (DMSO) has shown to recover cellular growth conditions brings new opportunities to exploit this host as , which was also confirmed by in silico analysis a platform for the industrial production of SA and other (Additional file 5). reduced by-products. Further validations included the comparison of in vivo In silico simulations were carried out using pFBA and measurements and in silico FVA predictions for mini- FVA methods predicting growth behaviour under chem- mum and maximum yields under various conditions ically defined medium. Model predictions were validated (Fig. 2). Despite being relatively variable (from 0.20 to − 1 using reported physiological data and flux distributions 0.23 g.g under glucose conditions), experimental from C experiments found in literature [32, 33, 51]. values for biomass yields were used to validate model The iBP722 model supports growth predictions on 19 predictions. Variations in experimental conditions, par- carbon sources under anaerobic conditions (Additional ticularly associated with culture media that is often sup- file 3), including C6 and C5 sugars that were found to plemented with yeast extract, interfering with carbon Pereira et al. BMC Systems Biology (2018) 12:61 Page 9 of 12 ab c Fig. 4 FVA analysis of A. succinogenes metabolism under varying growth conditions. Minimum and maximum FVA yields were computed − 1 maintaining 99% of the maximum growth rate and are given in mol.mol . (a) Predicted FVA yields (minimum-maximum) for key metabolic reactions under D-sorbitol growth conditions (considering PEP:PTS transport mechanism, Sorb_PTS) and glucose conditions (considering both PEP:PTS and symport transport mechanisms, Glc_PTS/symport, or only PEP:PTS transport, Glc_PTS). (b) Predicted FVA yields for SA production when changing ME2 reversibility, both under D-sorbitol or D-glucose conditions (considering only PEP:PTS transport). (c) Predicted FVA yields and pFBA yields (grey dots) for SA, FA and EtOH production under D-glucose conditions when changing CO availability (100% availability means unconstrained CO uptake, while 0% indicates a zero CO uptake flux value) 2 2 yields, or the initial sugar concentration that may affect experimental measurements. However, when using data bacterial growth due to substrate inhibition , may from C-labeling experiments  to constrain fluxes, explain these differences. Nevertheless, biomass compos- in particular the CO uptake rate (Fig. 3c), model predic- ition represented in the model is based on the work of tions improved, especially for reactions associated with McKinlay and co-workers , which is expected to pro- FA accumulation (PFL, FDHmq and EX_FA). vide accurate in silico predictions, especially regarding Overall, the iBP722 model allow us to evaluate the metabolic requirements to generate biomass contents production of SA, showing that maximum theoretical per unit of substrate (i.e. biomass yields, Y ). Additional yield for SA (1.1 g per g of glucose, assuming a symport x/s reaction constraints were included in the model to im- system and no ATP requirements for maintenance) is prove model predictions. For instance, flux constraints comparable to those predicted using E. coli or S. cerevi- of L-glutamate uptake and succinyl-CoA synthase were siae models (1.1 and 0.8 g per g of glucose for anaerobic changed to improve carbon-to-nitrogen ratios according conditions using iJO1366  and iMM904 , re- to experimental measurements . Moreover, the ener- spectively, under the same previous assumptions). Thus, getic requirement for non-growth associated maintenance, metabolic capabilities seem equivalent to other organ- i.e. the amount of ATP spent for cellular maintenance isms being exploited for SA production. It also allows without growth, was adjusted to improve predicted bio- describing the impact of growth conditions on the pro- mass yields. duction of C3 and C4 fermentative by-products. The Fermentative products ratios were also compared, carbon split between C4 and C3 pathways has been in- showing some consistency between experimental mea- vestigated and showed to be influenced by various fac- surements and FVA predictions. Predicted mass ratios tors like the available reducing power (i.e. NADH/NAD for SA/AA and SA/FA are higher than for SA/FA, which ratio) or CO availability, therefore affecting the SA pro- is consistent with in vivo data; but FVA spans for SA/ duction in A. succinogenes growing cultures [64, 69]. In AA and SA/FA suggest a huge flexibility in these ratios. silico predicted SA yields are higher with more reduced This suggests that carbon flux distributions in the C3 carbon sources (e.g. minimum and maximum FVA yields − 1 branch are rather challenging to predict. In fact, FVA of 0.44 and 0.46 g.g on sorbitol compared to 0.41 and − 1 flux spans of fermentative pathways (Fig. 3b) indicate a 0.44 g.g on glucose, correspondingly, assuming the high metabolic flexibility in the accumulation of same sugar transport mechanism) and higher CO avail- by-products which, in some cases are inconsistent with ability, favouring carbon flux through the C4 branch, as Pereira et al. BMC Systems Biology (2018) 12:61 Page 10 of 12 a consequence of higher reducing power and higher PPCK of C5 and C6 sugars, as well as other low-cost carbon carboxylation activity. Moreover, flexibility in sugar trans- sources (e.g. glycerol or lactose) and its metabolic flexi- port mechanisms or enzymes reversibility (e.g. ME2) may bility may provide some advantages over other lead to increased levels of SA, as carbon flux partitioning SA-producing strains, like recombinant Escherichia coli between C3 and C4 pathways was shown to be largely af- or Mannheimia succiniciproducens . fected (Fig. 4). The reverse activity of ME2 has shown to increase SA yields, redirecting part of the carbon flux Additional files from the C3 toward the C4 branch through the carboxyl- ation of pyruvate to L-malate, with the simultaneous pro- Additional file 1: Biomass composition of A. succinogenes 130Z. (DOCX 43 kb) duction of reducing power (NADPH). On the other hand, Additional file 2: Details on the iBP722 model. (DOCX 24 kb) transport activities limited to PEP:PTS-based systems de- Additional file 3: Testing A. succinogenes growth on different carbon crease SA yields, since PEP is used as the energy source sources. (DOCX 22 kb) for sugar uptake generating pyruvate, which is necessarily Additional file 4: Details on transport mechanisms included in the consumed via the C3 branch. iBP722 model. (DOCX 37 kb) Additional file 5: Growth predictions on glycerol with and without Conclusions DMSO. (DOCX 21 kb) In this work, the GSMM of A. succinogenes 130Z (iBP722) was reconstructed and validated using different sets of ex- Abbreviations Metabolites perimental data from literature. The reconstruction of the AA: Acetic acid; AA : Extracellular acetic acid; ACCOA: Acetyl-coenzyme A; ex model included the compilation of functionally annotated E4P: Erythrose-4-phosphate; EtOH: Ethanol; EtOH : Extracellular ethanol; ex metabolic genes and the corresponding coding proteins, F6P: Fructose-6-phosphate; FA: Formic acid; FA : Extracellular formic acid; ex FUM: Fumarate; FUM : Extracellular fumarate; G3P: Glyceraldehyde-3- ex as well as associated biochemical reactions. The model phosphate; G6P: Glucose-6-phosphate; Glc : Extracellular glucose; ex was complemented with a biomass equation and spontan- MAL: Malate; OAA: Oxaloacetate; PEP: Phosphoenolpyruvate; PYR: Pyruvate; eous and transport reactions. It was further amended after R5P: Ribose-5-phosphate; S7P: Sedoheptulose-7-phosphate; SA: Succinic acid.; SA : Extracellular succinic acid ex a gap filling process, including the identification of genetic evidences based on homology searches. Model accuracy to Reactions predict growth phenotypes and the production of fermen- ACKr: Acetate kinase; ALCD2x: Alcohol dehydrogenase; ENO: Enolase; EX_CO : Carbon dioxide exchange reaction; EX_AA: Acetic acid exchange tative by-products was evaluated using FVA and pFBA reaction; EX_EtOH: Ethanol exchange reaction; EX_FA: Formic acid exchange simulation methods. The ability to predict changes in car- reaction; EX_FUM: Fumarate exchange reaction; EX_Glc: Glucose exchange bon flux distributions due to environmental perturbations reaction; EX_SA: Succinic acid exchange reaction; FDHmq: Formate dehydrogenase; FRD2: Fumarate reductase; FUM: Fumarase; like CO limitations or alterations in the redox state was G6PDH2r: Glucose 6-phosphate dehydrogenase of the oxidative pentose also tested. Predicted SA yields were in good agreement phosphate pathway; ME2: Malic enzyme; OAADC: Oxaloacetate with experimental data, suggesting that the model is able decarboxylase; PFK: Phosphofructokinase; PFL: Pyruvate formate-lyase; PGI: Glucose-6-phosphate isomerase; PPCK: PEP carboxykinase; PYK: Pyruvate to characterize the fermentative metabolism under various kinase conditions. The increase in CO availability showed to have a positive impact in SA yields, which is consistent Acknowledgements with reported data [48, 49]. As such, optimal conditions The authors want to thank Apostolis Koutinas and co-workers from Agricultural University of Athens for the helpful discussions. for increased SA yields may include increased CO avail- ability, the use of more reduced carbon sources like sorb- Funding itol or the use of external energy source like H . Financially supported by BRIGIT (KBBE-2012-6-311935, FP7 project Contract nr Besides improving process conditions for the production 311935) and by the Portuguese Foundation for Science and Technology (FCT) under the scope of the strategic funding of UID/BIO/04469/2013 unit of SA, the design of microbial strains by metabolic en- and COMPETE 2020 (POCI-01-0145-FEDER-006684), in addition to the gineering to increase the flux through the C4 branch has BioTecNorte operation (NORTE-01-0145-FEDER-000004) funded by European been attempted, albeit with limited success . Model Regional Development Fund under the scope of Norte2020 - Programa Operacional Regional do Norte. predictions show that changes in sugar transport systems, CO fixation activities or the reversibility of the malic Availability of data and materials enzyme may have an impact in the SA yield. Therefore, it The GSMM generated during the current study is available at http:// darwin.di.uminho.pt/models and at BioModels with the identifier is expected that modifications in the metabolism that MODEL1804130001. A. succinogenes 130Z complete genome - GenBank would include these activities would further improve the accession number NC_009655. SA yield. Overall, the iBP722 model enables a better under- Authors’ contributions Model development: SC, BP, JM, PV; Design of simulation scenarios: PV, SS, IR; standing of the metabolic behaviour and capabilities of Model simulations and data analysis SC, BP, PV; Paper writing: SC, BP; this organism, which can be explored to further improve Contribution to paper writing: PV, SS, IR. All authors read and approved the SA productivity. The capacity to consume a wide range final version of the manuscript. Pereira et al. BMC Systems Biology (2018) 12:61 Page 11 of 12 Ethics approval and consent to participate 16. Zeikus JG, Jain MK, Elankovan P. Biotechnology of succinic acid production Not applicable. and markets for derived industrial products. Appl Microbiol Biotechnol. 1999;51:545–52. 17. Kamzolova SV, Vinokurova NG, Shemshura ON, Bekmakhanova NE, Lunina Competing interests JN, Samoilenko VA, Morgunov IG. The production of succinic acid by yeast The authors declare that they have no competing interests Yarrowia lipolytica through a two-step process. Appl Microbiol Biotechnol. 2014;98:7959–69. 18. Raab AM, Gebhardt G, Bolotina N, Weuster-Botz D, Lang C. Metabolic Publisher’sNote engineering of Saccharomyces cerevisiae for the biotechnological production Springer Nature remains neutral with regard to jurisdictional claims in of succinic acid. Metab Eng. 2010;12:518–25. published maps and institutional affiliations. 19. Yan D, Wang C, Zhou J, Liu Y, Yang M, Xing J. Construction of reductive pathway in Saccharomyces cerevisiae for effective succinic acid fermentation Author details at low pH value. Bioresour Technol. 2014;156:232–9. 1 2 SilicoLife Lda, Rua do Canastreiro 15, 4715-387 Braga, Portugal. CEB - 20. Liu Y, Wu H, Li Q, Tang X, Li Z, Ye Q. Process development of succinic acid Centre of Biological Engineering, University of Minho, Campus de Gualtar, production by Escherichia coli NZN111 using acetate as an aerobic carbon 4710-057 Braga, Portugal. Instituto de Tecnologia Química e Biológica source. Enzym Microb Technol. 2011;49:459–64. António Xavier, Universidade Nova de Lisboa (ITQB-NOVA), Oeiras, Portugal. 21. Scholten E, Renz T, Thomas J. Continuous cultivation approach for fermentative succinic acid production from crude glycerol by Basfia Received: 21 December 2017 Accepted: 14 May 2018 succiniciproducens DD1. Biotechnol Lett. 2009;31:1947–51. 22. Kim DY, Yim SC, Lee PC, Lee WG, Lee SY, Chang HN. Batch and continuous fermentation of succinic acid from wood hydrolysate by Mannheimia succiniciproducens MBEL55E. In Enzyme Microb Technol Volume. 2004;35:648–53. References 23. Lee SY, Kim JM, Song H, Lee JW, Kim TY, Jang Y-S. From genome sequence 1. Van der Werf MJ, Guettler MV, Jain MK, Zeikus JG. Environmental and to integrated bioprocess for succinic acid production by Mannheimia physiological factors affecting the succinate product ratio during succiniciproducens. Appl Microbiol Biotechnol. 2008;79:11–22. carbohydrate fermentation by Actinobacillus sp. 130Z. Arch Microbiol. 1997; 24. Myriant [http://www.myriant.com]. 167:332–42. 25. BASF [http://www.basf.com]. 2. Liu YP, Zheng P, Sun ZH, Ni Y, Dong JJ, Zhu LL. Economical succinic acid 26. BioAmber [http://www.bio-amber.com]. production from cane molasses by Actinobacillus succinogenes. Bioresour 27. Cok B, Tsiropoulos I, Roes AL, Patel MK. Succinic acid production derived from Technol. 2008;99:1736–42. carbohydrates: an energy and greenhouse gas assessment of a platform 3. Urbance SE, Pometto AL, DiSpirito A a., Denli Y: Evaluation of succinic acid chemical toward a bio-based economy. Biofuels Bioprod Biorefin. 2014;8:16–29. continuous and repeat-batch biofilm fermentation by Actinobacillus 28. Choi S, Song CW, Shin JH, Lee SY. Biorefineries for the production of top succinogenes using plastic composite support bioreactors. Appl Microbiol building block chemicals and their derivatives. Metab Eng. 2015:223–39. Biotechnol 2004, 65:664–670. 29. Jansen MLA, van Gulik WM. Towards large scale fermentative production of 4. Guettler MV, Rumler D, Jainf MK. Actinobacillus succinogenes sp. nov., a novel succinic acid. Curr Opin Biotechnol. 2014;30:190–7. succinic-acid-producing strain from the bovine rumen. Int J Syst Bacteriol. 30. Lee SJ, Song H, Lee SY. Genome-based metabolic engineering of 1999;49(1 999):207–16. Mannheimia succiniciproducens for succinic acid production. Appl Environ 5. Gunnarsson IB, Alvarado-morales M, Angelidaki I. Utilization of CO2 fixating Microbiol. 2006;72:1939–48. bacterium Actinobacillus succinogenes 130Z for simultaneous biogas 31. Yan Q, Zheng P, Dong JJ, Sun ZH. A fibrous bed bioreactor to improve the upgrading and biosuccinic acid production. Environ Sci Technol. 2014;48: productivity of succinic acid by Actinobacillus succinogenes. J Chem Technol 12464–8. Biotechnol. 2014;89:1760–6. 6. Shen N, Qin Y, Wang Q, Liao S, Zhu J, Zhu Q, Mi H, Adhikari B, Wei Y, Huang 32. McKinlay JB, Zeikus JG, Vieille C. Insights into Actinobacillus succinogenes R. Production of succinic acid from sugarcane molasses supplemented with fermentative metabolism in a chemically defined growth medium. Appl a mixture of corn steep liquor powder and peanut meal as nitrogen Environ Microbiol. 2005;71:6651–6. sources by Actinobacillus succinogenes. Lett Appl Microbiol. 2015;60:544–51. 33. McKinlay JB, Vieille C. 13C-metabolic flux analysis of Actinobacillus 7. Zheng P, Dong JJ, Sun ZH, Ni Y, Fang L. Fermentative production of succinogenes fermentative metabolism at different NaHCO3 and H2 succinic acid from straw hydrolysate by Actinobacillus succinogenes. concentrations. Metab Eng. 2008;10:55–68. Bioresour Technol. 2009;100:2425–9. 34. Li J, Jiang M, Chen K-QQ, Ye Q, Shang L-AA, Wei P, Ying H-JJ, Chang H-NN. 8. Wan C, Li Y, Shahbazi A, Xiu S. Succinic acid production from cheese whey Effect of redox potential regulation on succinic acid production by using Actinobacillus succinogenes 130 Z. Appl Biochem Biotechnol. 2008;145: Actinobacillus succinogenes. Bioprocess Biosyst Eng. 2010;33:911–20. 111–9. 35. Carvalho M, Matos M, Roca C, Reis M a M. Succinic acid production from 9. Chen K, Jiang M, Wei P, Yao J, Wu H. Succinic acid production from acid glycerol by Actinobacillus succinogenes using dimethylsulfoxide as electron hydrolysate of corn fiber by Actinobacillus succinogenes. Appl Biochem acceptor. New Biotechnol. 2014;31:133–9. Biotechnol. 2010;160:477–85. 36. van Heerden CD, Nicol W. Continuous succinic acid fermentation by 10. Jiang M, Chen K, Liu Z, Wei P, Ying H, Chang H. Succinic acid production by Actinobacillus succinogenes. Biochem Eng J. 2013;73:5–11. Actinobacillus succinogenes using spent brewer’s yeast hydrolysate as a 37. Otero JM, Cimini D, Patil KR, Poulsen SG, Olsson L, Nielsen J. Industrial nitrogen source. Appl Biochem Biotechnol. 2010;160:244–54. systems biology of Saccharomyces cerevisiae enables novel succinic acid cell 11. Xi YL, Chen KQ, Dai WY, Ma JF, Zhang M, Jiang M, Wei P, Ouyang PK. factory. PLoS One. 2013;8 Succinic acid production by Actinobacillus succinogenes NJ113 using corn 38. Sajo Mienda B, Shahir Shamsir M. Model-driven in silico glpC gene knockout steep liquor powder as nitrogen source. Bioresour Technol. 2013;136:775–9. predicts increased succinate production from glycerol in Escherichia coli. 12. Li Q, Yang M, Wang D, Li W, Wu Y, Zhang Y, Xing J, Su Z. Efficient AIMS Bioeng. 2015;2:40–8. conversion of crop stalk wastes into succinic acid production by 39. Xu G, Zou W, Chen X, Xu N, Liu L, Chen J. Fumaric acid production in Actinobacillus succinogenes. Bioresour Technol. 2010;101:3292–4. Saccharomyces cerevisiae by in silico aided metabolic engineering. PLoS One. 13. Chen K, Zhang H, Miao Y, Jiang M, Chen J. Succinic acid production from 2012;7 enzymatic hydrolysate of sake lees using Actinobacillus succinogenes 130Z. 40. Raab AM, Lang C. Oxidative versus reductive succinic acid production in the Enzym Microb Technol. 2010;47:236–40. 14. Chen KQ, Li J, Ma JF, Jiang M, Wei P, Liu ZM, Ying HJ. Succinic acid yeast Saccharomyces cerevisiae. Bioeng Bugs. 2011;2 production by Actinobacillus succinogenes using hydrolysates of spent yeast 41. Cheng KK, Wang GY, Zeng J, Zhang JA. Improved succinate production by cells and corn fiber. Bioresour Technol. 2011;102:1704–8. metabolic engineering. Biomed Res Int. 2013;2013 15. Yu J, Li Z, Ye Q, Yang Y, Chen S. Development of succinic acid production 42. Lin H, Bennett GN, San KY. Fed-batch culture of a metabolically engineered from corncob hydrolysate by Actinobacillus succinogenes. J Ind Microbiol Escherichia coli strain designed for high-level succinate production and yield Biotechnol. 2010;37:1033–40. under aerobic conditions. Biotechnol Bioeng. 2005;90:775–9. Pereira et al. BMC Systems Biology (2018) 12:61 Page 12 of 12 43. Liu R, Liang L, Chen K, Ma J, Jiang M, Wei P, Ouyang P. Fermentation of 68. Mo ML, Palsson BO, Herrgård MJ. Connecting extracellular metabolomic xylose to succinate by enhancement of ATP supply in metabolically measurements to intracellular flux states in yeast. BMC Syst Biol. 2009;3:37. engineered Escherichia coli. Appl Microbiol Biotechnol. 2012;94:959–68. 69. Pateraki C, Patsalou M, Vlysidis A, Kopsahelis N, Webb C, Koutinas AA, 44. Zhang X, Jantama K, Moore JC, Jarboe LR, Shanmugam KT, Ingram LO. Koutinas M. Actinobacillus succinogenes: advances on succinic acid Metabolic evolution of energy-conserving pathways for succinate production and prospects for development of integrated biorefineries. production in Escherichia coli. Proc Natl Acad Sci U S A. 2009;106:20180–5. Biochem Eng J. 2016;112:285–303. 70. Guarnieri MT, Chou YC, Salvachúa D, Mohagheghi A, St. John PC, Peterson 45. ya LL, ming LR, feng MJ, quan CK, Jiang M, Wei P. Increased production of DJ, Bomble YJ, Beckham GT. Metabolic engineering of Actinobacillus succinic acid in Escherichia coli by overexpression of malate dehydrogenase. succinogenes provides insights into succinic acid biosynthesis. Appl Environ Biotechnol Lett. 2011;33:2439–44. Microbiol. 2017;83 46. Yang J, Wang Z, Zhu N, Wang B, Chen T, Zhao X. Metabolic engineering of 71. Song H, Lee JW, Choi S, You JK, Hong WH, Lee SY. Effects of dissolved CO2 Escherichia coli and in silico comparing of carboxylation pathways for high levels on the growth of Mannheimia succiniciproducens and succinic acid succinate productivity under aerobic conditions. Microbiol Res. 2014;169: production. Biotechnol Bioeng. 2007;98:1296–304. 432–40. 72. Brink HG, Nicol W. Succinic acid production with Actinobacillus succinogenes: 47. Jiang M, Ma J, Wu M, Liu R, Liang L, Xin F, Zhang W, Jia H, Dong W. rate and yield analysis of chemostat and biofilm cultures. Microb Cell Progress of succinic acid production from renewable resources: metabolic Factories. 2014;13:111. and fermentative strategies. Bioresour Technol. 2017; 73. Lin H, Bennett GN, San KY. Metabolic engineering of aerobic succinate 48. Zou W, Zhu L-W, Li H-M, Tang Y-J. Significance of CO2 donor on the production systems in Escherichia coli to improve process productivity and production of succinic acid by Actinobacillus succinogenes ATCC 55618. achieve the maximum theoretical succinate yield. Metab Eng. 2005;7:116–27. Microb Cell Factories. 2011;10:87. 49. Xi Y, Chen K, Li J, Fang X, Zheng X, Sui S, Jiang M, Wei P. Optimization of culture conditions in CO2 fixation for succinic acid production using Actinobacillus succinogenes. J Ind Microbiol Biotechnol. 2011;38:1605–12. 50. McKinlay JB, Laivenieks M, Schindler BD, McKinlay A a, Siddaramappa S, Challacombe JF, Lowry SR, Clum A, Lapidus AL, Burkhart KB, Harkins V, Vieille C. A genomic perspective on the potential of Actinobacillus succinogenes for industrial succinate production. BMC Genomics 2010, 11:680. 51. McKinlay JB, Shachar-Hill Y, Zeikus JG, Vieille C. Determining Actinobacillus succinogenes metabolic pathways and fluxes by NMR and GC-MS analyses of 13C-labeled metabolic product isotopomers. Metab Eng. 2007;9:177–92. 52. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403–10. 53. Eddy SR. Profile hidden Markov models. Bioinformatics. 1998;14(9):755–63. 54. The UniProt Consortium. UniProt: the universal protein knowledgebase. Nucleic Acids Res. 2018;46(5):2699. 55. Dias O, Rocha M, Ferreira EC, Rocha I. Reconstructing genome-scale metabolic models with merlin. Nucleic Acids Research. 2015;43(8):3899–910. 56. Kanehisa M, Goto S. KEGG: Kyoto Encyclopaedia of genes and genomes. Nucl Acids Res. 2000;28:27–30. 57. Saier MH, Reddy VS, Tamang DG, Västermark Å. The transporter classification database. Nucleic Acids Res. 2014;42 58. Caspi R, Billington R, Ferrer L, Foerster H, Fulcher CA, Keseler IM, Kothari A, Krummenacker M, Latendresse M, Mueller LA, Ong Q, Paley S, Subhraveti P, Weaver DS, Karp PD. The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases. Nucleic Acids Res. 2016;44:D471–80. 59. Flamholz A, Noor E, Bar-even A, Milo R. EQuilibrator - the biochemical thermodynamics calculator. Nucleic Acids Res. 2012;40 60. Santos S, Rocha I. A computation tool for the estimation of biomass composition from genomic and transcriptomic information. Adv. Intell. Syst. Comput. 2016:161–9. 61. Juty N, Ali R, Glont M, Keating S, Rodriguez N, Swat MJ, Wimalaratne SM, Hermjakob H, Le Novère N, Laibe C, Chelliah V. BioModels: content, features, functionality, and use. CPT Pharmacometrics Syst Pharmacol. 2015:55–68. 62. Rocha I, Maia P, Evangelista P, Vilaça P, Soares S, Pinto JP, Nielsen J, Patil KR, Ferreira EC, Rocha M. OptFlux: an open-source software platform for in silico metabolic engineering. BMC Syst Biol. 2010;4:45. 63. Bradfield MFA, Nicol W. Continuous succinic acid production from xylose by Actinobacillus succinogenes. Bioprocess Biosyst Eng. 2016;39:233–44. 64. Li J, Jiang M, Chen K, Shang L, Wei P, Ying H, Ye Q, Ouyang P, Chang H. Enhanced production of succinic acid by Actinobacillus succinogenes with reductive carbon source. Process Biochem. 2010;45:980–5. 65. Schindler BD, Joshi RV, Vieille C. Respiratory glycerol metabolism of Actinobacillus succinogenes 130Z for succinate production. J Ind Microbiol Biotechnol. 2014;41:1339–52. 66. Lin SKC, Du C, Koutinas A, Wang R, Webb C. Substrate and product inhibition kinetics in succinic acid production by Actinobacillus succinogenes. Biochem Eng J. 2008;41:128–35. 67. Orth JD, Conrad TM, Na J, Lerman JA, Nam H, Feist AM, Palsson BØ. A comprehensive genome-scale reconstruction of Escherichia coli metabolism—2011. Mol Syst Biol. 2011;
BMC Systems Biology – Springer Journals
Published: May 30, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera