Update mitochondrial reaction bounds based on Gibbs free energy and MitoCore comparison

[hchapman1]

Background

As part of the mitochondrial curation on branch feat/curateMitochondria, I compared the mitochondrial compartment in Human-GEM to the MitoCore model. MitoCore is a manually curated model of human cental metabolism that focuses on mitochondrial pathways (DOI 10.1186/s12918-017-0500-7, link).
During this comparison I identified a set of mitochondrial reactions where the reversibility in Human-GEM disagreed with the corresponding reaction in MitoCore.

To investigate which directionality is more plausible, I evaluated these reactions using eQuilibrator (https://equilibrator.weizmann.ac.il/), and then proposed updated bounds in Human-GEM that are consistent with both the sign/magnitude of the standard transformed Gibbs free energy (ΔG°′), and the direction reported in MitoCore.

For each reaction that disagreed with reversibility in MitoCore I retrieved ΔG°′ (kJ/mol) from eQuilibrator and applied the following threshold to decide reversibility:

• If −30 kJ/mol ≤ ΔG°′ ≤ +30 kJ/mol → reaction treated as reversible.
• If ΔG°′ < −30 kJ/mol → constrained to be forward-only (lb = 0, ub = 1000).
• If ΔG°′ > +30 kJ/mol → constrained to be reverse-only (lb = -1000, ub = 0).

Proposed changes

The table below summarises the affected reactions, their ΔG°′ values from eQuilibrator, and the proposed new bounds.

HumanGEMID	Equation (Human-GEM / BiGG-style)	ΔG°′ (kJ/mol)	current_lb	current_ub	new_lb	new_ub
MAR02148	nadp + occoa <=> MAM00059m + h + nadph	56	-1000	1000	-1000	0
MAR02149	ddcacoa + nadp <=> dd2coa + h + nadph	56	-1000	1000	-1000	0
MAR02183	tdccoa + nadp <=> MAM00066m + h + nadph	56	-1000	1000	-1000	0
MAR02187	dcacoa + nadp <=> dc2coa + h + nadph	56	-1000	1000	-1000	0
MAR04220	hLkynr + h2o + nadp <=> h + Lkynr + nadph + o2	422	-1000	1000	-1000	0
MAR03809	citr__L + h + pi <=> cbp + orn	30.3	-1000	1000	-1000	0
MAR03139	3oddcoa + coa --> accoa + dcacoa	30.2	0	1000	-1000	0
MAR02207	MAM00053m + h + nadph <=> MAM02122m + nadp	-56	-1000	1000	0	1000
MAR03806	glu__L + 2.0 h + nadh <=> h2o + glu5sa + nad	-50.9	-1000	1000	0	1000
MAR00479	dhap + h + nadh --> nad + glyc3p	-22.6	0	1000	-1000	1000
MAR03136	dd2coa + h2o --> MAM00174m	0.4	0	1000	-1000	1000
MAR03137	MAM00174m + nad --> 3oddcoa + h + nadh	15	0	1000	-1000	1000
MAR03143	dc2coa + h2o --> MAM00181m	0.4	0	1000	-1000	1000
MAR03144	MAM00181m + nad --> 3odccoa + h + nadh	15	0	1000	-1000	1000
MAR03150	MAM00059m + h2o --> MAM00183m	0.4	0	1000	-1000	1000
MAR03151	MAM00183m + nad --> MAM00892m + h + nadh	15	0	1000	-1000	1000
MAR03157	MAM00053m + h2o --> MAM00182m	-0.7	0	1000	-1000	1000
MAR03158	MAM00182m + nad --> MAM00882m + h + nadh	16	0	1000	-1000	1000
MAR03164	b2coa + h2o --> 3hbcoa	-3.3	0	1000	-1000	1000
MAR03166	3hbcoa + nad --> aacoa + h + nadh	13.8	0	1000	-1000	1000
MAR03213	mmcoa__S --> mmcoa__R	0.2	0	1000	-1000	1000
MAR03215	mmcoa__R --> succoa	-8.4	0	1000	-1000	1000
MAR03744	akg + val__L --> 3mob + glu__L	1.7	0	1000	-1000	1000
MAR03753	h2o + 2mp2coa --> 3hibutcoa	5.6	0	1000	-1000	1000
MAR03757	3hmp + nad --> 2mop + h + nadh	18	0	1000	-1000	1000
MAR03765	akg + leu__L --> 4mop + glu__L	0.3	0	1000	-1000	1000
MAR03778	akg + ile__L --> 3mop + glu__L	-5.9	0	1000	-1000	1000
MAR03787	acac + succoa --> aacoa + succ	12.6	0	1000	-1000	1000
MAR03861	3sala + akg + h --> 3snpyr + glu__L	-9.3	0	1000	-1000	1000
MAR03879	hcys__L + ser__L --> h2o + cyst__L	-15	0	1000	-1000	1000
MAR03897	glu5p + h + nadph --> glu5sa + nadp + pi	-24	0	1000	-1000	1000
MAR04012	adn + atp --> adp + amp + h	-15	0	1000	-1000	1000
MAR04288	akg + h + lys__L + nadph --> h2o + nadp + saccrp__L	-16	0	1000	-1000	1000
MAR04340	atp + for + thf --> 10fthf + adp + pi	-2	0	1000	-1000	1000
MAR04474	6pgc + nadp --> co2 + nadph + ru5p__D	-6.8	0	1000	-1000	1000
MAR04656	fol + nadph --> dhf + nadp	-10	0	1000	-1000	1000
MAR04667	h2o + nad + saccrp__L --> L2aadp6sa + glu__L + h + nadh	16	0	1000	-1000	1000
MAR06768	akg + tyr__L --> 34hpp + glu__L	0.4	0	1000	-1000	1000
MAR04190	arg__L + h + nadph + o2 <=> h2o + nwharg + nadp	N/A	-1000	1000	0	1000

eQuilibrator could not return a ΔG° For MAR04190 due to missing data on metabolites, but MitoCore provided evidence for directionality for both reactions. For MAR04190 (L-arginine + O₂ + NADPH + H⁺ → Nω-hydroxy-L-arginine + NADP⁺ + H₂O), MitoCore lists the directionality evidence as “Oxygen consumption, BRENDA, KEGG, HumanCyc”. The database entries all describe nitric-oxide synthase (EC 1.14.13.39) as an NADPH- and oxygen-dependent monooxygenase acting from L-arginine towards Nω-hydroxy-L-arginine / citrulline and NO (https://www.brenda-enzymes.org/enzyme.php?ecno=1.14.13.39 ).

Updated model (on branch feat/curateMitochondria)

[JonathanRob]

Hi @hchapman1, really nice work - thank you for this.

If you haven't already, I would recommend checking to see if any of these changes impair any essential pathway(s), or allow for physiologically infeasible solutions. For example, by using checkTasks with the VerifyModel or Full task lists.

[edkerk]

Due to a bug in RAVEN, the new Human-GEM.yml has mistakenly lost all rxnFrom and metFrom annotations, I'm fixing this bug right now.

[hchapman1]

I tested the updated model using RAVEN’s checkTasks for the following metabolic task sets: Essential tasks, Verification tasks and Full task list

With the reversibility changes applied, the model failed a subset of these tasks. When I made either MAR03213 or MAR03215 irreversible, all Essential, Verification, and Full tasks passed.

I chose to keep MAR03215 irreversible, since its |ΔG°′| from eQuilibrator is larger than that of MAR03213, making irreversible directionality more consistent with the thermodynamic evidence.

[JonathanRob]

Ok that's great! It's probably fine if not all tasks in the Full tasks list pass, since from a recent issue it sounded like some of them may anyway be incorrectly defined. But if the updated model passes all the essential and verification tasks, then that is very reassuring that these changes could be good to implement.