Reconstruction and analysis of genome-scale metabolic model of a photosynthetic bacterium
© Montagud et al; licensee BioMed Central Ltd. 2010
Received: 5 February 2010
Accepted: 17 November 2010
Published: 17 November 2010
Synechocystis sp. PCC6803 is a cyanobacterium considered as a candidate photo-biological production platform - an attractive cell factory capable of using CO2 and light as carbon and energy source, respectively. In order to enable efficient use of metabolic potential of Synechocystis sp. PCC6803, it is of importance to develop tools for uncovering stoichiometric and regulatory principles in the Synechocystis metabolic network.
We report the most comprehensive metabolic model of Synechocystis sp. PCC6803 available, i Syn669, which includes 882 reactions, associated with 669 genes, and 790 metabolites. The model includes a detailed biomass equation which encompasses elementary building blocks that are needed for cell growth, as well as a detailed stoichiometric representation of photosynthesis. We demonstrate applicability of i Syn669 for stoichiometric analysis by simulating three physiologically relevant growth conditions of Synechocystis sp. PCC6803, and through in silico metabolic engineering simulations that allowed identification of a set of gene knock-out candidates towards enhanced succinate production. Gene essentiality and hydrogen production potential have also been assessed. Furthermore, i Syn669 was used as a transcriptomic data integration scaffold and thereby we found metabolic hot-spots around which gene regulation is dominant during light-shifting growth regimes.
i Syn669 provides a platform for facilitating the development of cyanobacteria as microbial cell factories.
Cyanobacteria, which have been model organisms since the early 70s of the past century , are a widespread group of photoautotrophic microorganisms, which originated, evolved, and diversified early in Earth's history . It is commonly accepted that cyanobacteria played a crucial role in the Precambrian phase by contributing oxygen to the atmosphere . All cyanobacteria combine the ability to perform an oxygenic photosynthesis (resembling that of chloroplasts) with typical prokaryotic features, like performing anoxygenic photosynthesis by using hydrogen sulfide (H2S) as the electron donor or fixing atmospheric dinitrogen (N2) into ammonia (NH3). Relevance of this phylum covers from evolutionary studies  to biotechnological applications, including biofuel production . Synechocystis sp. PCC6803 is a cyanobacterium that is considered as a good candidate for developing a photo-biological cell factory towards production of a variety of molecules of socio-economic interest, with CO2 (and/or sugars) as carbon source and light (and/or sugars) as energy source . The diversity of potential applications in this sense is broad. Works have been published on heterologous production of metabolites such as isoprene , poly-beta-hydroxybutyrate , biofuels  and bio-hydrogen [9, 10] - an energy vector of global interest .
Synechocystis sp. PCC6803 is capable of growing under three different growth conditions as marked by the utilized carbon source (/s) . This causes that three distinct modes of operation are interweaved over the same metabolic network, viz., i) photoautotrophy, where energy comes from light and carbon from CO2; ii) heterotrophy, where energy and carbon source is a saccharide, for instance glucose; and, iii) mixotrophy, a combination of the above two, where light is present as well as a combination of two carbon sources: glucose and CO2. Reconstruction of a genome-scale metabolic model for this model photo-synthetic bacterium is one of the main goals of the current study. Genome-scale metabolic network reconstruction is, in essence, a systematic assembly and organization of all the reactions which build up the metabolism of a given organism; and has been of great interest in the post-genomic era. The variety of applications of such a metabolic model  includes the possibility of assessing projects for the production and optimization of an added value metabolite. If a model is formulated properly, it is expected to allow simulating environmental and genetic perturbations in the metabolic network. Thus, together with appropriate constraints, a metabolic model would partially represent a virtual organism - an in silico model that allows probing possible flux distributions inside the cell under different environmental conditions and for a given genetic make-up. Towards this end, a variety of tools/algorithms are available , including flux balance analysis (FBA) [15, 16], minimization of metabolic adjustments (MOMA) , regulatory on-off minimization (ROOM)  and metabolic control analysis (MCA) [19, 20].
Synechocystis sp. PCC6803 genome was sequenced, annotated and made publicly available in 1996 [21, 22] and has been the target of some metabolic modeling effort, especially for central carbon metabolic reconstructions [23, 24]. The work from Yang et al focused on a metabolic model of glycolysis, tricarboxylic acid cycle and pentose phosphate pathway that was simulated under heterotrophic and mixotrophic conditions. Shastri and Morgan  studied a metabolic model with the same pathways under autotrophic conditions and compared their results to the ones from Yang et al. These two works considered one lumped reaction for the photosynthesis of the system. More recently, an uncurated reaction list with a biomass composition represented by central carbon metabolites has been published . This model, however, is not suitable for simulations due to lack of proper biomass equation, lumped nature of some key reactions and missing reactions.
The large quantity of information featured in public databases, like details about genomes , pathways , enzymes  or proteins  can be used from different databases to gather all published data for one specific organism. However, the lack of quality must be considered as a major drawback of some of the databases: false positives, false negatives as well as wrongly annotated objects may hinder efforts of collecting accurate data . Consequently, manual reconstruction by detailed inspection of each and every reaction, biomass equation based on metabolic building blocks (such as amino acids and nucleotides), consistency and integrity of the network is a pre-requisite for creating a high quality and useful metabolic model . The current study presents such manually curated reconstruction for Synechocystis sp. PCC6803 and demonstrates some of its potential applications.
The present model features a detailed biomass equation which encompasses all the building blocks that are needed for a flux distribution simulation that reflects observed phenotype. No lumped reactions are present and photosynthesis is described as a set of 19 reactions, thus enabling the tracing of the corresponding fluxes. Furthermore, different analyses are performed by using this metabolic reconstruction, including reaction knock-out simulations, flux variability analysis and identification of transcriptional regulatory hotspots. Overall, i Syn669 is a valuable tool towards the development of a photo-biological production platform. The model will also contribute to the existing set of genome-scale models with a virtue of being one of the first stoichiometric models that account for photosynthesis.
Results and Discussion
Genome-scale metabolic network reconstruction
A complete literature examination, including databases, biochemistry textbooks and the annotated genome sequence, was needed in order to extract the current state of the art on known metabolic reactions within the metabolic network of Synechocystis sp. PCC6803. For a thorough overview of the process of metabolic model reconstruction, refer to very instructive work by Forster et al as well as review by Feist et al. In detail, the reconstruction started with the annotation and genomic sequence files of Synechocystis sp. PCC6803 [21, 22]. These files were used with Pathway Tools software  in order to build a database of all the genes, proteins and metabolites presents in the organism. The list of reactions was then retrieved from Pathway Tools; EC numbers and stoichiometry of the reactions were checked and verified with the help of the Enzyme nomenclature database  and KEGG pathway database . Reactions were elementally balanced except for protons, so that chemical conversions were coherent. In some of the reactions present in these databases, metabolites were reported in a non-specific form (e.g. 'an alcohol'). This is insufficient for metabolic model simulation and, so, corresponding organism-specific metabolites had to be identified . Additionally, in a large number of reactions cofactors were not completely clarified: an enzyme being capable of using NADH or NADPH or both. In the latter, two reactions were included in the reconstructed metabolic network. Determination of reversibility of the reactions was assisted by specific enzyme databases, like BRENDA . If no conclusive evidence was reported, reactions were set to be reversible.
Distribution of the model reactions as per cognate genes
Number of reactions
-With assigned genes
-With no cognate gene
·EC reactions not annotated
·Needed for biomass simulation
The product of this reconstruction process was a set of reactions that encompass all the known metabolite conversions that take place in Synechocystis sp. PCC6803. The resulting network, i Syn669, consists of 882 metabolic reactions and 790 metabolites (see Table 1 for more information). A total of 669 genes were included, to which 639 reactions were assigned (see Additional file 1 for details); the difference between the number of genes and assigned reactions is due to the presence of considerable number of protein complexes (e.g. photosynthetic or respiratory activities) and isoenzymes. Reactions with no cognate genes are also present in i Syn669, 20 passive transport reactions and 47 chemical conversions (not mediated by enzymes) were included. Additionally, a total of 79 reactions were included on the basis of biochemical evidence or physiological considerations, but currently with no annotated Open Reading Frame (ORF). i Syn669 genome-scale metabolic model is available in Additional file 2 (in OptGene  format).
i Syn669 spans all the biologically relevant flux nodes in the Synechocystis metabolism. Pyruvate, phosphoenolpyruvate (PEP), 3-phosphoglycerate, erythrose-4-phosphate and 2-oxoglutarate are main flux nodes for amino acids biosynthesis. Acetyl-CoA is an important flux node for fatty acids production, with high relevance for metabolic engineering towards biofuel production. Biosynthesis of nucleic acids comes from different metabolites, namely, ribose-5-phosphate, 5-phospho-beta-D-ribosyl-amine, L-histidine and L-glutamine. Moreover, with the information publicly available on databases, we can conclude that Synechocystis sp. PCC6803 bears an incomplete tricarboxylic acid cycle (TCA cycle), as it lacks 2-ketoglutarate dehydrogenase (EC 184.108.40.206). It has been published that glyoxylate shunt completes this cycle , permitting the recycling of TCA metabolites. Alternatively, aspartate transaminase (reaction 220.127.116.11a in i Syn669) can interconvert 2-ketoglutarate and oxaloacetate, thus bridging the gap of 2-ketoglutarate dehydrogenase, but short-circuiting TCA cycle.
Most connected metabolites in the iSyn669 metabolic network.
in E. coli
Simulations of the three metabolic modes
iSyn669 Biomass composition.
Growth under pure heterotrophy, or dark heterotrophy (in the absence of light) is a subject under study [42, 43], being the regular experimental design to give a short light pulse prior to the pure heterotrophic phase (light-activated heterotrophy). Nevertheless, the theoretical flux distribution under heterotrophic conditions is interesting by itself - especially in comparison with the flux distribution in a light-fed energy metabolism. Moreover, fluxes in the heterotrophy mode may help in obtaining insight into the variations under the mixotrophic condition, which is of high relevance for industrial applications .
Comparison of selected fluxes across different growth conditions.
beta-D-glucose + ATP → beta-D-glucose-6-phosphate + ADP
malate ↔ fumarate + H2O
D-ribose-5-phosphate ↔ D-ribulose-5-phosphate
PSII* + UQ + 2 H+ → PSII + UQH2
NADH + UQ + 7 H+ → NAD+ + UQH2 + 4 H+_peribac
3 H+_peribac + phosphate O4P + ADP ↔ 3 H+ + H2O + ATP
coenzyme A + acetate + ATP ↔ acetyl-CoA + diphosphate + AMP
Heterotrophy was simulated by considering glucose as the sole carbon source with uptake rate of 0.567 mmol gDW-1 h-1, entering the system through glcP glucose transporter (reaction TRANS-RXN59G-152 in i Syn669). With the purpose of having a pure heterotrophic state, photon uptake rate was constrained to 0; this caused photosynthesis fluxes to be shut down. In this case, glucose will be the source for the formation of carbon backbones for the building blocks of the cell, depicted in the biomass equation. The glycolytic and the oxidative mode of the pentose phosphate pathway were found to be active. Oxidative pentose phosphate pathway is the major pathway for glucose catabolism as was reported in reference . PEP carboxylase (reaction 18.104.22.168 in i Syn669) is the main anaplerotic flux to the TCA cycle. Carbon fixation efficiency is around 60%, the rest being released in the form of CO2, as reported in our previous work .
In contrast to dark heterotrophy, if a light-activated heterotrophy simulation is run, light enters the system and RuBisCO enzyme is active (reaction 22.214.171.124), fixing all the CO2 that was released in dark heterotrophy, boosting carbon efficiency to a theoretical 100%. In this case, global flux distribution as well as flux ranges resemble that of autotrophy more than that of the dark heterotrophy. Carbon skeletons are still produced through glycolysis and NAD(P)H is reduced along the glycolysis, pyruvate metabolism and TCA cycle. On the other hand, pentose phosphate pathway has shifted to the reductive mode due to RuBisCO activation and the corresponding flux is increased in magnitude. Carbon fixation happens at the RuBisCO level, thereby assimilating the CO2 produced by the glucose metabolism, and the production of ATP and NADPH through photosynthesis relieves the oxidative phosphorylation from draining NADPH to generate ATP.
Photoautotrophy was initially simulated considering an illumination of 0.15 mE m-2 s-1. Assuming that the mass of a typical Synechocystis sp. PCC6803 cell is 0.5 pg  and its radius is 1.75 μm , we estimated that the theoretical maximum illumination is 41563.26 mE gDW-1 h-1. An additional optimization step was performed in order to estimate physiologically meaningful photon uptake values that are closer to the experimental measurements . First, carbon uptake rate was found that resulted in a specific growth rate of 0.09 h-1, while the light intake was unconstrained. Next, the growth rate was constrained to this value and the second optimization problem was solved where light uptake was minimized. This minimization resulted in photon uptake for photosystem I (reaction _lightI) and photosystem II (reaction _lightII) being 0.8 mE gDW-1 h-1. Carbon sources used in simulating photoautotrophy conditions were carbon dioxide and carbonic acid, and its entrance to the system was mediated by RuBisCO (reaction 126.96.36.199 in i Syn669) and carbonic anhydrase (reaction 188.8.131.52b) respectively. As i Syn669 biomass equation encompasses all essential metabolite precursors, these will be the sinks of our network, while photons, carbon dioxide and/or carbonic acid will be the sources. Thus autotrophic fluxes will flow in the gluconeogenic direction and through the Calvin cycle, which is the reductive mode of the pentose phosphate pathway. PEP carboxylase is the main anaplerotic flux to the TCA cycle and glyoxylate shunt is inactive.
Photons, carbon dioxide and glucose are independent feed fluxes in this simulation. These fluxes entered the system through the same reactions as described for the previous growth modes. Carbon source presents, in this case, one more degree of freedom than in the rest of the conditions. In order to keep a comparative criterion across conditions, we normalized CO2 and glucose inputs to the same carbon uptake flux as in the case of the autotrophy and the heterotrophy. Photon uptake rates were also normalized in a similar manner to match the autotrophic state. Having the same metabolic sinks as the two previous modes and the sources from the both of them, it is logical to think that the resulting flux distribution will be a mixture of the autotrophic and heterotrophic simulations. Indeed, we observed that the mixotrophic flux distribution lies in-between the previous two states, being a bit closer to the heterotrophy. Glycolysis is present and glyoxylate is shut down; an active photosynthesis is present, oxidative phosphorylation is less stressed than in heterotrophy as the energy can be produced from the photon uptake; and Calvin cycle is active, as carbon sources are CO2 and glucose.
Flux variability analysis
Gene/Reaction knock-out analysis
The comprehensive set of reconstructed biochemical equations of i Syn669 and FBA simulations enabled us to further analyze the characteristics and potential of the Synechocystis metabolic network. This can be oriented towards the study of the reactions (and thereby the corresponding genes) that are necessary for the growth, or to in silico metabolic engineering for identification of targets for maximization of a given metabolite of socio-economic interest.
Interestingly, if we compare the proportion of the essential genes under FBA simulation in the metabolic networks of E. coli (187 genes, 15% of the total)  and Saccharomyces cerevisiae (148, 10% of the total)  with i Syn669, we find that Synechocystis has a significantly larger fraction of metabolic genes whose deletion obliterates biomass formation (304 genes, 34% of the total). One possible explanation for the difference in the relative proportion of essential genes in these three organisms would be an incomplete/incorrect annotation of the genome of Synechocystis sp. PCC6803. For example, if only one of the isoenzymes corresponding to a reaction is annotated, the corresponding in silico knock-out will result in a false negative prediction. It is important to note that the computational predictions of gene essentiality based on FBA are highly dependent on the growth medium used for the simulations. Thus, the comparison across different species may not be straight-forward. Moreover, it is also possible that the natural growth conditions of Synechocystis may have dictated selection for a relatively high proportion of essential genes. Such hypotheses need careful consideration of several factors and are beyond the scope of this work.
Production of value-added compounds
Synechocystis sp. PCC6803 is considered as a candidate photobiological production platform - it can potentially produce molecules of interest by using CO2 and light . To this end, i Syn669 can be used to perform simulations, not only for assessing the feasibility of producing a given compound, but also to identify potential metabolic engineering targets towards improved productivity. For example, FBA simulations can help in estimating maximum theoretical yields for the products/intermediates of interest. A product of obvious interest is hydrogen. In our previous work , we have estimated maximum theoretical hydrogen production values that are far from the current state of experimental reports. In silico studies can direct the efforts and counsel the scientists towards a hydrogen producing cyanobacteria that could be of impact. i Syn669 predicts, in autotrophic conditions, a theoretical H2 evolution rate of 0.17 mmol gDW-1 h-1 obliterating biomass growth. Else, the stoichiometry permits the evolution of 0.156 mmol gDW-1 h-1 of hydrogen with a biomass growth of 10% of the wild type (0.007 mmol gDW-1 h-1).
Succinate is an important metabolite for its biotechnological applications as well as for being a metabolite that bridges the TCA cycle with the electron transfer chain. As an example of the usefulness of the present metabolic model we have designed an in silico metabolic engineering strategy to improve the production of succinate. The underlying idea is to design a succinate over-producing metabolic network (through reaction knock-out simulations), whereas the intracellular fluxes are distributed so as to maximize the biological objective function (e.g. growth) . To this end, OptGene algorithm  was used together with Minimization Of Metabolic Adjustment (MOMA)  as a biological objective function. MOMA has been reported to provide better description of flux distributions in mutants or under un-natural growth conditions as opposed to FBA. A design objective function which copes with the metabolite of interest, succinate, has been determined maintaining the biological objective function as the biomass formation.
OptGene simulations for single, double and triple knock-out strategies were performed to obtain solutions with improved succinate production, but without drastically diminishing the biomass production. We used mixotrophic conditions, for which wild type optimal growth rate was 0.17909 mmol gDW-1 h-1. The best single knock-out was found to be the mutant of pyruvate kinase (reaction 184.108.40.206c in i Syn669 and genes sll0587 and sll1275) that has a succinate evolution of 0.5695 mmol gDW-1 h-1 with a growth rate of 0.0714 mmol gDW-1 h-1. Blocking this reaction, preventing pyruvate and phosphoenolpyruvate from using GTP and GDP would drive a high increase in succinate production. The flux between pyruvate and phosphoenolpyruvate can still be accomplished with reactions 220.127.116.11a and 18.104.22.168, but using ATP and ADP as cofactors. Double deletion did not improve the results from the single knock-out strain, evolving the same succinate production with the same growth rate. The best triple knock-out was found to be the combination of pyruvate kinase (reaction 22.214.171.124c in i Syn669 and genes sll0018 and sll0587), fructose-bisphosphate aldolase (reaction 126.96.36.199b in i Syn669 and genes slr0943 and sll1275) and succinate dehydrogenase (reaction _188.8.131.52 in i Syn669 and genes sll0823, sll1625 and slr1233). This simulated strain has a succinate evolution of 0.6999 mmol gDW-1 h-1 with a growth rate of 0.0688 mmol gDW-1 h-1. This design combines the blocking of the oxidation of succinate on the electron chain transfer through succinate dehydrogenase with the prevention of using GTP between pyruvate and phosphoenolpyruvate and the lack of an aldolase needed in the reductive mode of the pentose phosphate pathway. This leads to a situation where flux is directed to TCA cycle in order to meet with an overproduction of succinate.
These studies on knock-outs are reaction centered, even though the in vivo knock-out building will ultimately be through gene manipulations. This is the reason underlying the fact that we found 184.108.40.206c knock-out as the best result. This design would hint at the idea of selection of a mutated pyruvate kinase protein specific for ATP cofactor. This may be difficult to achieve on the bench, but has high biotechnological expectations.
i Syn669 as a data integration scaffold
Apart from the flux simulations, another important problem in the field of metabolic systems biology that can be addressed by using reconstructed genome-scale models is the integration of the different genome-wide bio-molecular abundance datasets, i.e. omics datasets, such as transcriptome and metabolome. An example of algorithms for carrying out such an integrative analysis through the use of genome-scale metabolic networks is Reporter Features [48, 49]. Reporter algorithm allows integration of omics data with bio-molecular interaction networks, thereby allowing identification of cellular regulatory focal points (i.e. reporter features), for instance reporter metabolites as regulatory hubs in the metabolic network.
In this work, Reporter Features software was used to integrate transcriptional information over the reconstructed Synechocystis sp. PCC6803 network allowing us to infer regulatory principles underlying metabolic flux changes following shifts in growth mode. In particular, we analyzed the data from a work  that reports the transcriptional changes caused in Synechocystis sp. PCC6803 by shifts from darkness to illumination conditions and back. As it can be understood from the rationale beneath the metabolic capabilities of this cyanobacterium, the presence or absence of light drives big changes in the flux distribution through the network, as discussed in the previous sections. We have focused our study on the relationship between the transcription of Synechocystis sp. PCC6803 genes and the reactions of the metabolic network. Associations between genes and reactions were identified, listing all the genes that performed or were involved in a specific reaction. With this information and the metabolic model, Reporter Features analysis was carried out. In brief, the analysis helped to identify metabolites around which the transcriptional changes are significantly concentrated. These metabolites are termed reporter metabolites as they represent key regulatory nodes in the network.
Gill et al designed the experiment so that Synechocystis was grown to mid-exponential phase (A730 = 0.6 to 0.8). Then, the lights were extinguished and RNA samples were taken after 24 h in the dark (full dark). Illumination was then turned back on for 100 min (transient light), followed immediately by an additional 100 min in the dark (transient dark).
KEGG orthology groups for the metabolic genes altered with the light shift.
All time points
Dark to Light
Light to Dark
Amino Acid Metabolism
Metabolism of Cofactors and Vitamins
Biosynthesis of Secondary
Biosynthesis of Polyketides
and Nonribosomal Peptides
All time points
Reporter metabolites for the light shift experiment.
Number of neighbors
Number of neighbors
Number of neighbors
All time points
Dark to Light
Light to Dark
peptidylproline (omega = 180)
(E, E)-farnesyl diphosphate
peptidylproline (omega = 0)
By using the metabolic sub-network search algorithm, we found 212 genes that have their expression changed across the arrays and that have a relationship with the metabolites of i Syn669 network. Furthermore, 50 genes were identified that are strongly co-regulated all along the profile of the experiment (Additional File 7, section a). This set of genes is characterized in two groups. The first set consists of the genes from photosynthesis (93.85%) and oxidative phosphorylation (6.15%). The second set is representative of a variety of genes from different pathways such as amino acid metabolism (39%), carbohydrate metabolism (22%), nucleotide metabolism (13%), nitrogen metabolism (13%) and metabolism of cofactors (9%) that globally regulates the entire metabolic network (see Table 5 for further details).
It can be expected that an experimental design like the one we have based our work on, which combines a shift from dark to light with a shift back to darkness, will encompass an important part of the regulatory changes the cell is undergoing in its natural habitat. In a glucose-deficient environment, the presence or absence of light is the main condition around which the Synechocystis metabolism gravitates . Indeed, one of the co-regulated sets consists of the genes coding for the proteins that work on, and around, the thylakoid membrane, let it be photosynthesis or oxidative phosphorylation genes.
Dark to light
Next, we considered the arrays that represent the shift from darkness to light, the first three arrays (from "24 hours of darkness" array to "60 minutes of light" array). Reporter metabolites were found to be largely within the nucleotide and amino acid metabolism (Table 6b). Some cofactors were also identified as regulation hubs like tetrahydrofolate, thioredoxin and adenosylcobinamide.
Sub-network search yielded set of 247 genes that have their expression changed across the first three arrays and that are related with i Syn669 reactions. Furthermore, 84 genes were identified that are strongly co-regulated across the three arrays (Additional File 7, section b). This set of genes cover photosynthesis (25%), oxidative phosphorylation (24%), amino acid metabolism (11%), carbohydrate metabolism (11%), nucleotide metabolism (10%) and metabolism of cofactors (10%).
This set of data arrays are indeed a good example of a cell's metabolic machinery starting up. After a 24 hour period in darkness where cell density did not change (see Figure 1 in Gill et al), light enters the system and the cell starts to synthesize new bio-molecules, mostly nucleotides so it can copy its genetic material and amino acids to build up proteins.
Light to dark
Finally, we considered the arrays that represent the shift from light to dark, data from "90 minutes of light" array to "60 minutes of dark" array. Similar to the previous case study, reporter metabolites were found to be focused on the nucleotide and amino acid metabolism (Table 6c). Additionally, the presence of metabolite a 1,4-alpha-D-glucan_n and its cognate a 1,4-alpha-D-glucan_n1 also stands out as they are involved in carbon reserves catabolism and anabolism.
With the help of the sub-network search, 133 genes were identified as being significantly co-regulated across those three arrays (Additional File 7, section c). This set comprises of the genes from photosynthesis (34%), oxidative phosphorylation (26%), amino acid metabolism (12%), carbohydrate metabolism (12%), nucleotide metabolism (7.5%) and metabolism of cofactors (4.5%).
This last set of data array is a scenario where metabolism is being shut down, as a consequence of the darkness and lack of carbohydrate source. Without light, photosynthesis is blocked and carbon fixation is nearly obliterated. Cells strive to build up carbon reserves (hence the presence of a 1,4-alpha-D-glucan_n as a reporter metabolite) and oxidative phosphorylation is the main energy pathway that remains present. Regulation is centered on the energy metabolism shift (60% of the total co-regulated sub-network), withholding amino acids and nucleotide precursors and keeping the cofactors available in a low-profile metabolism.
We have successfully reconstructed a genome-scale metabolic network for Synechocystis sp. PCC6803, called i Syn669, which allows simulating production of all the metabolic precursors of the organism. The metabolic reconstruction represents an up-to-date database that encompasses all knowledge available in public databases, scientific publications and textbooks on the metabolism of this cyanobacterium.
From the annotation publicly available, our metabolic network includes 882 metabolic reactions and 790 metabolites, as well as the information from 669 genes that have some relationship with the metabolic reactions. This model is the most complete and comprehensive work for Synechocystis sp. PCC6803 to date, which has its potential as the photosynthetic model organism. Interestingly, the reconstruction identified 79 reactions that should be present in the metabolism but with no cognate gene discovered yet; this should direct experimental work at the discovery of these genes. Topological characteristics of the network resemble those of other reconstructed microbial metabolic networks and thus provide an additional input for the analysis of their structural and organizational properties from evolutionary perspective.
Applicability of i Syn669 metabolic model was demonstrated by using a variety of computational analyses. Flux balance analysis was applied in order to simulate the three physiologically important growth conditions of cyanobacteria, viz., heterotrophic, mixotrophic and autotrophic. Our metabolic model was capable of simulating the production of the monomers or building blocks that build up the cells, in the range that is in agreement with the reported growth experiments. Our photosynthetic metabolic model includes all of the central metabolic pathways that previous works [23–25] considered. Regarding the parts from our model that overlap with the previous works (part of the central carbon metabolism), the predictions for the flux directionality changes following light shift match between those models and i Syn669. In fact, i Syn669 expands the flux study to all the pathways described in the Synechocystis sp. PCC6803 genome annotation. Further work should be directed at the definition of a detailed and descriptive biomass cell composition, so as to have a better representation of the biomass equation for simulation purposes.
Single reaction/gene knock-out simulations revealed 311 genes that are essential for the survival. Bearing in mind the distance from the efforts taken in the annotation of the genome of the bacteria and yeast models to that of the cyanobacterium, our study shows that Synechocystis sp. PCC6803 has a larger fraction of genes that are essential for producing biomass, as opposed to Escherichia coli and Saccharomyces cerevisiae. Further investigation of the causes for this difference will be of definite interest in understanding the genome annotation and/or the evolution of the metabolic network of Synechocystis.
Evaluation of the theoretical potential of this organism to produce hydrogen was assessed, in support of the efforts directed to this direction from several groups and scientific council initiatives. Present hydrogen production projects are far from the theoretical potential, but efforts in this field can trigger a very significant increase of the present hydrogen evolution rates in Synechocystis sp. PCC6803 or other photobiological production platforms candidates, e.g. Chlamydomonas reinhardtii, Nostoc punctiforme and Synechococcus species.
Suitability of the presented model for performing in silico metabolic engineering analysis was demonstrated by using OptGene software framework. Furthermore, we also show that i Syn669 can be used as a scaffold to integrate network-wide omics data. As a case study, we identified key reporter metabolites around which regulation during light shifts is organized, as well as gene sub-networks that were co-regulated across the light conditions.
Altogether, the genome-scale metabolic network of Synechocystis sp. PCC6803 (i Syn669) will be a valuable tool for the applied and fundamental research of Synechocystis sp. PCC6803, as well as for the broad field of metabolic systems biology. i Syn669 represents an important step for the integration of tools and knowledge from different disciplines towards development of photo-biological cell factories.
Metabolic network reconstruction
Pathway Tools software  was used to construct a Synechocystis-specific database of genes, proteins, enzymes and metabolites. Synechocystis sp. PCC6803 genome and annotation files were downloaded from NCBI Entrez Genome repository as of date 10 of September of 2008 . Pathway tools retrieved a first version of the network, which had to be checked with different kinds of databases depending on the information they bear. Databases used towards this purpose included Enzyme nomenclature database , KEGG pathway database , BioCyc genome database , BRENDA Enzyme database  and UniProt protein database .
Parts that characterize Synechocystis network, like the incomplete TCA cycle [52, 53], the presence of the glyoxylate shunt , the interconnected photosynthesis and oxidative phosphorylation  or the cyclic and non-cyclic electron transport related to these latter processes [55–57], were accounted for in detail.
At the end of the reconstruction process, four kinds of relationships were present in the database: reaction with cognate genes, reactions that needed to be included in the model in order to have metabolic precursors in the network (with no assigned genes), non-enzymatic reactions that have no related gene, and genes described in the annotations but with no assigned function. For an overview of the underlying process, please refer to Fortser et al work on the reconstruction of Saccharomyces cerevisiae metabolic network.
Linear programming for Flux Balance Analysis
The details are described elsewhere, for example in Stephanopoulos et al. This model describes cellular behavior under pseudo steady-state conditions, where S is stoichiometric matrix that contains the stoichiometric coefficients corresponding to all internal (balanced) metabolites. v is flux vector that corresponds to the columns of S. Given a set of experimentally-driven constraints, former equation was solved by using linear programming, the approach known as flux balance analysis, or FBA .
Since the number of reactions is typically larger than the number of metabolites, the system becomes underdetermined. In order to obtain a feasible solution for the intracellular fluxes, an optimization criterion on metabolic balances has to be imposed. This can be formulated by maximizing one of the biochemical reactions, e.g. biomass equation, subject to the mass balance and the capacity constraints.
where v j is the rate of the j th reaction. The elements of the flux vector v were constrained for the definition of reversible and irreversible reactions, v j, rev and v j, irr , respectively. Additionally, two set of equations were established, ν j, const , constrained metabolic reactions, and ν j, uptake , uptake reactions, which were bound by experimentally determined values from the literature. Biomass synthesis was considered as a drain of precursors or building blocks into a hypothetical biomass component. Flux through biomass synthesis reaction, being the biomass formation rate, is directly related to growth of the modeled organism . Table 3 shows the biomass composition that was considered in the i Syn669 metabolic model.
Simulations were performed with the OptGene software . Some capacity constraints had to be added in order to have a feasible solution for the linear programming problem. As an example, maximum uptake rates were determined as follows: maximum glucose uptake rate under heterotrophic conditions was found to be 0.85 mmol glucose gDW-1 h-1. Maximum CO2 uptake rate was found to be 3.7 mmol CO2 gDW-1 h-1. Additionally, we fixed the maintenance requirement for the heterotrophic case to be 1.67 ATP moles per mole of glucose consumed as was determined by ref , and was maintained for autotrophic and mixotrophic growth.
Segre et al introduced the method of minimization of metabolic adjustment (MOMA) to better understand the flux states of mutants. MOMA is based on the same stoichiometric constraints as FBA, but relaxes the assumption of optimal growth flux for the mutants, testing the hypothesis that the corresponding flux distribution is better approximated by the flux minimal response to the perturbation than by the optimal one.
is minimized. For details, please address to Segre et al.
Reporter Features algorithm
Reporter Features software  works on three kinds of information - network, omics data and association between genes and the nodes in the network. We have used Reporter Features for a transcriptomic analysis, so our three files were p-values file, resulting from a Student t-test run on transcriptomic data, interaction file, where reactions are connected to the corresponding substrates and products, and association file, where gene are associated to reactions they are involved in, either by coding for the enzyme or by regulating the gene that codes for the enzyme.
dry cell weight
flux balance analysis
metabolic control analysis
minimization of metabolic adjustments
Open Reading Frame
regulatory on-off minimization of metabolic fluxes
Ribulose-1,5-bisphosphate carboxylase oxygenase
- TCA cycle:
tricarboxylic acid cycle
This work was financially supported by MICINN TIN2009-12359 project ArtBioCom, EU FP7-KBBE-2007 project TarPol (contract n°212894) and EU FP6-NEST-2005 project BioModularH2 (contract n° 043340). AM thanks to Generalitat Valenciana grant BFPI/2007/283 and EN to Ministerio de Educación y Ciencia de España through the program Juan de la Cierva.
- Allen MM, Smith AJ: Nitrogen chlorosis in blue-green algae. Arch Mikrobiol. 1969, 69: 114-120. 10.1007/BF00409755View ArticlePubMedGoogle Scholar
- Tamagnini P, Axelsson R, Lindberg P, Oxelfelt F, Wunschiers R, Lindblad P: Hydrogenases and hydrogen metabolism of cyanobacteria. Microbiol Mol Biol Rev. 2002, 66: 1-20. table of contents 10.1128/MMBR.66.1.1-20.2002PubMed CentralView ArticlePubMedGoogle Scholar
- Schopf J: The Fossil Record: Tracing the Roots of the Cyanobacterial Lineage. The ecology of cyanobacteria. Edited by: Whitton B, Potts M. 2000, 13-35. Dordrecht: Kluwer Academic PublishersGoogle Scholar
- Shi T, Falkowski PG: Genome evolution in cyanobacteria: the stable core and the variable shell. Proc Natl Acad Sci USA. 2008, 105: 2510-2515. 10.1073/pnas.0711165105PubMed CentralView ArticlePubMedGoogle Scholar
- Tamagnini P, Leitao E, Oliveira P, Ferreira D, Pinto F, Harris DJ, Heidorn T, Lindblad P: Cyanobacterial hydrogenases: diversity, regulation and applications. FEMS Microbiol Rev. 2007, 31: 692-720. 10.1111/j.1574-6976.2007.00085.xView ArticlePubMedGoogle Scholar
- Lindberg P, Park S, Melis A: Engineering a platform for photosynthetic isoprene production in cyanobacteria, using Synechocystis as the model organism. Metab Eng. 2010, 12: 70-79. 10.1016/j.ymben.2009.10.001View ArticlePubMedGoogle Scholar
- Wu GF, Wu QY, Shen ZY: Accumulation of poly-beta-hydroxybutyrate in cyanobacterium Synechocystis sp. PCC6803. Bioresour Technol. 2001, 76: 85-90. 10.1016/S0960-8524(00)00099-7View ArticlePubMedGoogle Scholar
- Liu X, Curtiss R: Nickel-inducible lysis system in Synechocystis sp. PCC 6803. Proc Natl Acad Sci USA. 2009, 106: 21550-21554. 10.1073/pnas.0911953106PubMed CentralView ArticlePubMedGoogle Scholar
- Navarro E, Montagud A, Fernández de Córdoba P, Urchueguía JF: Metabolic flux analysis of the hydrogen production potential in Synechocystis sp. PCC6803. Int J Hydrogen Energy. 2009, 34: 8828-8838. 10.1016/j.ijhydene.2009.08.036.View ArticleGoogle Scholar
- McHugh K: Hydrogen production methods. 2005, Alexandria, Virginia: MPR Associates, IncGoogle Scholar
- Turner J, Sverdrup G, Mann M, Maness P, Kroposki B, Ghirardi M, Evans R, Blake D: Renewable hydrogen production. International Journal Energy Research. 2008, 32: 379-407. 10.1002/er.1372.View ArticleGoogle Scholar
- Herrero A, Flores E: The cyanobacteria: molecular biology, genomics, and evolution. 2008, Norfolk, UK: Caister Academic PressGoogle Scholar
- Oberhardt MA, Palsson BO, Papin JA: Applications of genome-scale metabolic reconstructions. Mol Syst Biol. 2009, 5: 320- 10.1038/msb.2009.77PubMed CentralView ArticlePubMedGoogle Scholar
- Patil KR, Akesson M, Nielsen J: Use of genome-scale microbial models for metabolic engineering. Curr Opin Biotechnol. 2004, 15: 64-69. 10.1016/j.copbio.2003.11.003View ArticlePubMedGoogle Scholar
- Varma A, Palsson BO: Metabolic capabilities of Escherichia coli: II. Optimal growth patterns. J Theor Biol. 1993, 165: 503-522. 10.1006/jtbi.1993.1203.View ArticleGoogle Scholar
- Edwards J, Ramakrishna R, Schilling C, Palsson B: Metabolic flux balance analysis. Metabolic engineering. Edited by: Lee S, Papoutsakis E. 1999, New York: Marcel Dekker IncGoogle Scholar
- Segre D, Vitkup D, Church GM: Analysis of optimality in natural and perturbed metabolic networks. Proc Natl Acad Sci USA. 2002, 99: 15112-15117. 10.1073/pnas.232349399PubMed CentralView ArticlePubMedGoogle Scholar
- Shlomi T, Berkman O, Ruppin E: Regulatory on/off minimization of metabolic flux changes after genetic perturbations. Proc Natl Acad Sci USA. 2005, 102: 7695-7700. 10.1073/pnas.0406346102PubMed CentralView ArticlePubMedGoogle Scholar
- Rapoport TA, Heinrich R, Jacobasch G, Rapoport S: A linear steady-state treatment of enzymatic chains. A mathematical model of glycolysis of human erythrocytes. Eur J Biochem. 1974, 42: 107-120. 10.1111/j.1432-1033.1974.tb03320.xView ArticlePubMedGoogle Scholar
- Kacser H, Burns JA: The control of flux. Symp Soc Exp Biol. 1973, 27: 65-104.PubMedGoogle Scholar
- Kaneko T, Sato S, Kotani H, Tanaka A, Asamizu E, Nakamura Y, Miyajima N, Hirosawa M, Sugiura M, Sasamoto S, et al.: Sequence analysis of the genome of the unicellular cyanobacterium Synechocystis sp. strain PCC6803. II. Sequence determination of the entire genome and assignment of potential protein-coding regions (supplement). DNA Res. 1996, 3: 185-209. 10.1093/dnares/3.3.185View ArticlePubMedGoogle Scholar
- Kaneko T, Nakamura Y, Sasamoto S, Watanabe A, Kohara M, Matsumoto M, Shimpo S, Yamada M, Tabata S: Structural analysis of four large plasmids harboring in a unicellular cyanobacterium, Synechocystis sp. PCC 6803. DNA Res. 2003, 10: 221-228. 10.1093/dnares/10.5.221View ArticlePubMedGoogle Scholar
- Yang C, Hua Q, Shimizu K: Metabolic flux analysis in Synechocystis using isotope distribution from 13C-labeled glucose. Metab Eng. 2002, 4: 202-216. 10.1006/mben.2002.0226View ArticlePubMedGoogle Scholar
- Shastri AA, Morgan JA: Flux balance analysis of photoautotrophic metabolism. Biotechnol Prog. 2005, 21: 1617-1626. 10.1021/bp050246dView ArticlePubMedGoogle Scholar
- Fu P: Genome-scale modeling of Synechocystis sp. PCC6803 and prediction of pathway insertion. Journal of Chemical Technology & Biotechnology. 2009, 84: 473-483.View ArticleGoogle Scholar
- Karp PD, Ouzounis CA, Moore-Kochlacs C, Goldovsky L, Kaipa P, Ahren D, Tsoka S, Darzentas N, Kunin V, Lopez-Bigas N: Expansion of the BioCyc collection of pathway/genome databases to 160 genomes. Nucleic Acids Res. 2005, 33: 6083-6089. 10.1093/nar/gki892PubMed CentralView ArticlePubMedGoogle Scholar
- Kanehisa M, Araki M, Goto S, Hattori M, Hirakawa M, Itoh M, Katayama T, Kawashima S, Okuda S, Tokimatsu T, Yamanishi Y: KEGG for linking genomes to life and the environment. Nucleic Acids Res. 2008, 36: D480-484. 10.1093/nar/gkm882PubMed CentralView ArticlePubMedGoogle Scholar
- Chang A, Scheer M, Grote A, Schomburg I, Schomburg D: BRENDA, AMENDA and FRENDA the enzyme information system: new content and tools in 2009. Nucleic Acids Res. 2009, 37: D588-592. 10.1093/nar/gkn820PubMed CentralView ArticlePubMedGoogle Scholar
- The universal protein resource (UniProt). Nucleic Acids Res. 2008, 36: D190-195.Google Scholar
- Weise S, Grosse I, Klukas C, Koschutzki D, Scholz U, Schreiber F, Junker BH: Meta-All: a system for managing metabolic pathway information. BMC Bioinformatics. 2006, 7: 465- 10.1186/1471-2105-7-465PubMed CentralView ArticlePubMedGoogle Scholar
- Feist AM, Herrgard MJ, Thiele I, Reed JL, Palsson BO: Reconstruction of biochemical networks in microorganisms. Nat Rev Microbiol. 2009, 7: 129-143.PubMed CentralView ArticlePubMedGoogle Scholar
- Forster J, Famili I, Fu P, Palsson BO, Nielsen J: Genome-scale reconstruction of the Saccharomyces cerevisiae metabolic network. Genome Res. 2003, 13: 244-253. 10.1101/gr.234503PubMed CentralView ArticlePubMedGoogle Scholar
- Karp PD, Paley S, Romero P: The Pathway Tools software. Bioinformatics. 2002, 18 (Suppl 1): S225-232.View ArticlePubMedGoogle Scholar
- Bairoch A: The ENZYME database in 2000. Nucleic Acids Res. 2000, 28: 304-305. 10.1093/nar/28.1.304PubMed CentralView ArticlePubMedGoogle Scholar
- Yang C, Hua Q, Shimizu K: Quantitative analysis of intracellular metabolic fluxes using GC-MS and two-dimensional NMR spectroscopy. J Biosci Bioeng. 2002, 93: 78-87.View ArticlePubMedGoogle Scholar
- Pearce J, Carr NG: The metabolism of acetate by the blue-green algae, Anabaena variabilis and Anacystis nidulans. J Gen Microbiol. 1967, 49: 301-313.View ArticlePubMedGoogle Scholar
- Patil KR, Rocha I, Forster J, Nielsen J: Evolutionary programming as a platform for in silico metabolic engineering. BMC Bioinformatics. 2005, 6: 308- 10.1186/1471-2105-6-308PubMed CentralView ArticlePubMedGoogle Scholar
- Feist AM, Henry CS, Reed JL, Krummenacker M, Joyce AR, Karp PD, Broadbelt LJ, Hatzimanikatis V, Palsson BO: A genome-scale metabolic reconstruction for Escherichia coli K-12 MG1655 that accounts for 1260 ORFs and thermodynamic information. Mol Syst Biol. 2007, 3: 121- 10.1038/msb4100155PubMed CentralView ArticlePubMedGoogle Scholar
- Zelezniak A, Pers TH, Soares S, Patti ME, Patil KR: Metabolic network topology reveals transcriptional regulatory signatures of type 2 diabetes. PLoS Comput Biol. 2010, 6: e1000729- 10.1371/journal.pcbi.1000729PubMed CentralView ArticlePubMedGoogle Scholar
- Stephanopoulos G, Aristidou AA, Nielsen JH: Metabolic engineering: principles and methodologies. 1998, San Diego: Academic PressGoogle Scholar
- Schuetz R, Kuepfer L, Sauer U: Systematic evaluation of objective functions for predicting intracellular fluxes in Escherichia coli. Mol Syst Biol. 2007, 3: 119- 10.1038/msb4100162PubMed CentralView ArticlePubMedGoogle Scholar
- Anderson SL, McIntosh L: Light-activated heterotrophic growth of the cyanobacterium Synechocystis sp. strain PCC 6803: a blue-light-requiring process. J Bacteriol. 1991, 173: 2761-2767.PubMed CentralPubMedGoogle Scholar
- Carr NG, Whitton BA: The Biology of cyanobacteria. 1982, Berkeley: University of California PressGoogle Scholar
- Pelroy RA, Rippka R, Stanier RY: Metabolism of glucose by unicellular blue-green algae. Arch Mikrobiol. 1972, 87: 303-322. 10.1007/BF00409131View ArticlePubMedGoogle Scholar
- Loferer-Krossbacher M, Klima J, Psenner R: Determination of bacterial cell dry mass by transmission electron microscopy and densitometric image analysis. Appl Environ Microbiol. 1998, 64: 688-694.PubMed CentralPubMedGoogle Scholar
- Lawrence BA, Suarez C, DePina A, Click E, Kolodny NH, Allen MM: Two internal pools of soluble polyphosphate in the cyanobacterium Synechocystis sp. strain PCC 6308: an in vivo 31P NMR spectroscopic study. Arch Microbiol. 1998, 169: 195-200. 10.1007/s002030050560View ArticlePubMedGoogle Scholar
- Stephanopoulos G, Alper H, Moxley J: Exploiting biological complexity for strain improvement through systems biology. Nat Biotechnol. 2004, 22: 1261-1267. 10.1038/nbt1016View ArticlePubMedGoogle Scholar
- Oliveira AP, Patil KR, Nielsen J: Architecture of transcriptional regulatory circuits is knitted over the topology of bio-molecular interaction networks. BMC Syst Biol. 2008, 2: 17- 10.1186/1752-0509-2-17PubMed CentralView ArticlePubMedGoogle Scholar
- Patil KR, Nielsen J: Uncovering transcriptional regulation of metabolism by using metabolic network topology. Proc Natl Acad Sci USA. 2005, 102: 2685-2689. 10.1073/pnas.0406811102PubMed CentralView ArticlePubMedGoogle Scholar
- Gill RT, Katsoulakis E, Schmitt W, Taroncher-Oldenburg G, Misra J, Stephanopoulos G: Genome-wide dynamic transcriptional profiling of the light-to-dark transition in Synechocystis sp. strain PCC 6803. J Bacteriol. 2002, 184: 3671-3681. 10.1128/JB.184.13.3671-3681.2002PubMed CentralView ArticlePubMedGoogle Scholar
- NCBI Entrez Genome for Synechocystis sp. PCC6803. http://www.ncbi.nlm.nih.gov/sites/entrez?Db=genome&Cmd=ShowDetailView&TermToSearch=112
- Pearce J, Leach CK, Carr NG: The incomplete tricarboxylic acid cycle in the blue-green alga Anabaena variabilis. J Gen Microbiol. 1969, 55: 371-378.View ArticlePubMedGoogle Scholar
- Vazquez-Bermudez MF, Herrero A, Flores E: Uptake of 2-oxoglutarate in Synechococcus strains transformed with the Escherichia coli kgtP gene. J Bacteriol. 2000, 182: 211-215. 10.1128/JB.182.1.211-215.2000PubMed CentralView ArticlePubMedGoogle Scholar
- Peschek GA, Löffelhardt W, Schmetterer G: The phototrophic prokaryotes. 1999, New York: Kluwer Academic/PlenumView ArticleGoogle Scholar
- Rubio FC, Camacho FG, Sevilla JM, Chisti Y, Grima EM: A mechanistic model of photosynthesis in microalgae. Biotechnol Bioeng. 2003, 81: 459-473. 10.1002/bit.10492View ArticlePubMedGoogle Scholar
- Albertsson P: A quantitative model of the domain structure of the photosynthetic membrane. Trends Plant Sci. 2001, 6: 349-358. 10.1016/S1360-1385(01)02021-0View ArticlePubMedGoogle Scholar
- Allen J: Photosynthesis of ATP-electrons, proton pumps, rotors, and poise. Cell. 2002, 110: 273-276. 10.1016/S0092-8674(02)00870-XView ArticlePubMedGoogle Scholar
- Herdman M, Janvier M, Waterbury J, Rippka R, Stanier R: Deoxyribonucleic Acid Base Composition of Cyanobacteria. Journal of General Microbiology. 1979, 111: 63-71.View ArticleGoogle Scholar
- Tasaka Y, Gombos Z, Nishiyama Y, Mohanty P, Ohba T, Ohki K, Murata N: Targeted mutagenesis of acyl-lipid desaturases in Synechocystis: evidence for the important roles of polyunsaturated membrane lipids in growth, respiration and photosynthesis. EMBO J. 1996, 15: 391-396.Google Scholar
- Miao X, Wu Q, Wu G, Zhao N: Changes in photosynthesis and pigmentation in an agp deletion mutant of the cyanobacterium Synechocystis sp. Biotechnol Lett. 2003, 25: 391-396. 10.1023/A:1022446330284View ArticlePubMedGoogle Scholar
- Burrows EH, Chaplen FW, Ely RL: Optimization of media nutrient composition for increased photofermentative hydrogen production by Synechocystis sp. PCC6803. International Journal of Hydrogen Energy. 2008, 33: 6092-6099. 10.1016/j.ijhydene.2008.07.102.View ArticleGoogle Scholar