TM4. T00023_518_C5. 7. SEED: 2-oxoglutarate transport in via proton symport. iIT341: 2-oxoglutarate reversible transport via symport. TM88. T00023_2580_C5.
H. pylori model data and validation Models comparison merlin was used to perform the re-annotation of the genome (1) and for developing a novel GSMM for Helicobater pylori, the iTR383 model (Resende, T., Correia, D.M.M. and Rocha, I. (2015) Reconstruction and validation of a genome-scale metabolic model for Helicobacter pylori 26695. In preparation), comprising 640 reactions. The comparison of this model with a previously published model, the iIT341 (2), comprising 477 reactions and a model exported from the modelSEED (720 reactions) platform is presented next. For this, all modelSEED and iIT341 metabolite identifiers were converted into KEGG identifiers using an internally developed database.
D C B
iTR 230 20 98 15
29
A
A – Full consensus B – Consensus (ignore compartment) C – Consensus (ignore proton) D – Consensus (ignore compartment and proton)
49
103
iIT341 107
0 13 0
50
0 13 0
SEED 410
Figure S3.1. Modified Venn diagram of the comparison of the reactions available in the three Helicobacter pylori models. The intersection areas represent the number of reactions common between the respective models. The reactions present in the consensus areas are clustered as follows: A - “Full consensus” for a perfect match; B - “Consensus (ignore compartment)” if reactions only match when the compartments are ignored; C – “Consensus (ignore proton)” for reactions that match if protons are ignored; D – “Consensus (ignore compartment and proton)” if reactions are only a match if protons and compartments are ignored.
The comparison of the reactions available in the three models is illustrated in the modified Venn diagram shown in Figure S3.1. The areas in which the models intersect, the consensus areas, represent the number of reactions in common between the respective models. However, the reactions present in the consensus areas are further clustered according to the type of similarity. If the reactions are a perfect match they are allocated to the A - “Full consensus” set; the B - “Consensus (ignore compartment)” cluster includes reactions that only match if the compartments are ignored; the protonation state of some metabolites might differ between models, thus some reactions, though being similar, may entail protons in their stoichiometry. Hence, clusters C – “Consensus (ignore proton)” and D – “Consensus
(ignore compartment and proton)” encompass reactions that match if protons are ignored and if protons and compartments are ignored, respectively. Ignoring compartments leads to duplicate matches between the iTR383 and the other models (iIT341 and modelSEED). This occurs because the iTR383 model has to transport metabolites from the exterior to the periplasm and then to the cytosol, whilst there is a direct transport (exterior to cytosol) in the models with just two compartments. As shown in Figure S3.1, all three models have 236 reactions (223 for iIT341 and modelSEED) in common. The difference between the numbers of common reactions is explained due to the duplicate matches when ignoring compartments. Though 103 reactions are exactly the same in all models, ignoring compartments allows increasing this number by 15 reactions (11 for iIT341 and modelSEED). When protons are ignored, yet accounting for compartmentalization, 98 identical reactions previously unmatched are detected. Finally, ignoring compartmentalization and protons in the remaining reactions allows identifying 20 additional reactions in iTR383 that are matched by 11 reactions in the other models. The list of duplicate reactions for the intersection of all three models is presented in Table S3.1. Table S3.1. Equivalence of the one step transport reactions in iIT341 and modelSEED, and the two steps transport reactions in the iTR383 model, between the external compartment and cytosol. SEED / iIT341 1 2 3 4 5 6 7 8 9 10 11 12 13
SEED: CO2 transport via diffusion iIT341: CO2 transporter via diffusion SEED: O2 transport via diffusion iIT341: o2 transport (diffusion) SEED: H2Ot5 iIT341: H2O transport via diffusion SEED: Urea transport via facilitate diffusion iIT341: Urea transport via facilitate diffusion SEED: L-lactate reversible transport via proton symport iIT341: L-lactate reversible transport via proton symport SEED: Aspartate-H symport iIT341: L-aspartate reversible transport via proton symport SEED: 2-oxoglutarate transport in via proton symport iIT341: 2-oxoglutarate reversible transport via symport SEED: acetoacetate transport via proton symport iIT341: acetoacetate transport via proton symport SEED: D-alanine transport in via proton symport iIT341: D-Alanine transport via proton symport SEED: PIt6 iIT341: phosphate reversible transport via symport SEED: succinate transporter in/out via proton symport iIT341: succinate transport via proton symport SEED: L-malate transport in via proton symport iIT341: L-malate reversible transport via proton symport SEED: fumarate transport in/out via proton symport iIT341: fumarate transport out via proton antiport
iTR383 External membrane Plasma membrane TM90 TM91 TM26
TM75
TM89
T00078_C5
TM68
T00031_1107_C5
TM32
T00023_3590_C5
TM4
T00023_518_C5
TM88
T00023_2580_C5
TM53
T00023_715_C5
TM27
T01482_1_C5
TM21
T00023_2054_C5
TM35
T03413_2069_C5
TM33
T00023_3472_C5
TM34
T00023_2130_C5
The iTR383 has 95 (29 perfect + 20 w/o compartments + 29 w/o proton + 17 w/o both) reactions that match approximately 17% of the reactions (29 perfect + 16 w/o compartments + 29 w/o proton + 10 w/o both) in the iIT341 model and cannot be found in the modelSEED model. Again, the difference in the number of reactions is again explained by two steps transportation in iTR383 instead of the direct transportation in iIT341. Likewise, about 12% of the iTR383 reactions (49 perfect + 12 w/o compartments + 17 w/o proton + 1 w/o both) can be found on modelSEED and do not exist in iIT341.
Finally the modelSEED and the iIT341 models have 63 (50 perfect + 13 w/o proton) equal reactions that cannot be found in the iTR383 model. Likewise, the difference in the number of reactions is related to the number of compartments in the different models. The list of duplicate reactions for the intersection of the iTR383 and models iIT341 and modelSEED is presented in Table S3.2. Table S3.2. Equivalence of the one step transport reactions in iIT341 and modelSEED to the two steps transport reactions in the iTR383 model, between the external compartment and cytosol. model 1 2 3 4 5 6 7 8
iIT341
9 10 11 12 13 14
modelSEED
reaction Guanine transport H2CO3 transport formate transport via diffusion Hypoxanthine transport pyruvate reversible transport via proton symport L-alanine transport in via proton symport L-lysine reversible transport xanthine transport in via proton symport acetate reversible transport via proton symport L-threonine reversible transport via proton symport oroatp Potassium uptake Cobalt transport Magnesium transport
iTR383 External membrane Plasma membrane TM43 TM76 TM60 TM81 TM58 T00023_3561_C5 TM44 TM77 TM31 T00023_1609_C5 T01482_2_C5
TM1
TM12 TM50
T01650_1_C5 TM79
TM52
TM87
TM17
TM86
TM46 TM116 TM104 TM118
TM78 TM117 TM105 TM119
The iIT341 model has 63 reactions (50 perfect matches and 13 ignoring protons) in common with the modelSEED in silico cell. These models have the same number of compartments, thus ignoring compartments does not increase the number of reactions matched between the models. The iTR383 model has 230, the iIT341 has 107 and the modelSEED 410 reactions that cannot be found in the other models, establishing that all models are different having their own peculiarities.
Specific growth rate assessment using a defined medium with glutamate as carbon source The assessment of the specific growth rate was performed by running simulations using the same environmental conditions (Env 1), proposed in (3), in all models. However, as shown in Table S3.3, some adjustments were performed so that growth could be assessed on the modelSEED and the iIT341 models. The latter required a couple of additional metabolites so that growth could be attained, whilst the former required several more compounds (most of which were present on the reaction that represents the formation of biomass) on the environmental conditions for achieving growth. Table S3.3. Environmental conditions used to grow H. pylori in silico. (* - metabolites present in the biomass equation). name
Env 1
iIT341 essential metabolites
modelSEED essential metabolites
Metabolites C00025_L-Glutamate_C5H9NO4 C00001_H2O_H2O C00007_Oxygen_O2 C00009_Orthophosphate_H3PO4 C00011_CO2_CO2 C00059_Sulfate_H2SO4 C00262_Hypoxanthine_C5H4N4O C00378_Thiamine_C12H17N4OS C14818_Fe2+_Fe C00062_L-Arginine_C6H14N4O2 C00135_L-Histidine_C6H9N3O2 C00079_L-Phenylalanine_C9H11NO2 C00073_L-Methionine_C5H11NO2S C00183_L-Valine_C5H11NO2 C00123_L-Leucine_C6H13NO2 C00407_L-Isoleucine_C6H13NO2 L-Alanine exchange Pimelate exchange DNA_replication_C_00001* Protein_biosynthesis_C_00001* RNA_transcription_C_00001* Ala-Leu_b Ca2_b* Cl-_b* Co2_b* Cu2_b* fe3_b* gly-pro-L_b Gly-Tyr_b Inosine_b K_b* L-Serine_b* Mg_b* Mn2_b* Zn2_b* Fumarate_b
Lower Bound mmol.gDW.h-1 -0.9348 -10.0 -10.0 -10.0 -10.0 -10.0 -10.0 -10.0 -10.0 -6.9444 -0.6944 -0.2 -0.168 -0.1416 -0.1388 -0.0444 -0.1 -0.1 -1000.0 -1000.0 -1000.0 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1
Upper Bound mmol.gDW.h-1 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 0.0 0.0 0.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0 10000.0
Essentiality tests The essentiality tests performed by Thiele et al. (2) in the iIT341 model were also performed in the iTR383 model. Although five environmental conditions are reported in the aforementioned study, only two (rich and the min4) were used to evaluate the iTR383 model, as the in vivo essentiality tests were performed in rich media. The two media formulations were implemented in this study with the following adaptations. Protoheme was replaced by heme (as the former does not exist on the iTR383 model) and the metabolites listed next were excluded from the media formulations, since these metabolites are not available in the iTR383 model: pimelate, D-serine, cytidine, deoxyadenosine, deoxycytidine, deoxyuridine, thymidine, uracyl, acryladmide, acetaldehyde, acetamide, ethanol, galactose, H2, sodium and nitrate. These metabolites are not present in the iTR383 because there are no reactions in this model that encompass such metabolites. Moreover, concerning the model retrieved from the modelSEED, as shown in Table S3.3 by the large number of essential metabolites required, significant curation is still required, and thus essentiality tests were not performed in such model. As shown in Table S3.4 the iTR383 model attains better predictions on three occasions (cells coloured with a green shadow). The first consists of attaining growth when knocking out gene HP0476, on the Min 4 medium. The second is the lack of growth when eliminating gene HP0512. Lastly, the growth of an HP1257 defective mutant cannot be achieved on rich media, which is in agreement with the experimental data. The iIT341 has better predictions than iTR383 in one occasion (cell coloured with a red shadow). The HP0360 knockout in the former leads to consuming D-galactose as carbon source for growth because, although not associated to any gene, this model has a reaction for metabolising galactose (galactokinase). However, during the H. pylori re-annotation, no evidences of D-galactose metabolizing enzymes (galactokinase - 2.7.1.6) were found in H. pylori’s genome, and thus such reaction is not available in the iTR383. If the lack of this capability proves to be real, in this case, H. pylori’s growth might be associated to a nutrient present in the experimental rich medium that is not present in the in silico rich medium. Overall, it can be concluded that the iTR383 behaviour, when accessing model predictions to experimental data on essentiality, is slightly better than the iIT341 model. For more information about this test please refer to (Resende et al, 2015).
Table S3.4. Comparison of in silico predictions to in vivo data (adapted from (2)). Gene Locus HP0002, HP1574 HP0004, HP1186 HP0055 HP0072, HP0073 HP0071 HP0075 HP0086 HP0121 HP0191 to HP0193 HP0197 HP0212 HP0215 HP0255 HP0293 HP0329 HP0360 HP0372 HP0380 HP0381 HP0389 HP0476 HP0509 HP0512 HP0588 to HP0591 HP0687 HP0735 HP0740 HP0802 HP0804 HP0824, HP0825 HP0829 HP0832 HP0875 HP1011 HP1038 HP1050 HP1052 HP1084 HP1087 HP1091 HP1099 HP1100 HP1100 HP1108, HP1111 HP1108, HP1111 HP1180 HP1257 HP1385 HP1399 HP1418 HP1495 HP1505
Min4 − + + + + − + + + − − − + − − − + + − + + − − − − + − − − − + − + − − + − − − + + + + − − + − + + − + −
Rich − + + + + − + + + − − − + − + − + + + + + − − − + + − − − − + − + + − + − + − + + + + + + + − + + − + −
Experimental − + + + + − − + + + − + + + + + + + + + + − − − − − − − − − + + + − − + + − − + − − + − + + − − + − + −
References 1. Resende, T., Correia, D.M., Rocha, M. and Rocha, I. (2013) Re-annotation of the genome sequence of Helicobacter pylori 26695. J. Integr. Bioinform., 10, 233. 2. Thiele, I., Vo, T.D., Price, N.D. and Palsson, B.Ø. (2005) Expanded metabolic reconstruction of Helicobacter pylori (iIT341 GSM/GPR): an in silico genome-scale characterization of single- and double-deletion mutants. J. Bacteriol., 187, 5818–30. 3. Correia, D.M.M. (2014) Systems analysis of metabolism in Helicobacter pylori.
