β-Oxidation of free fatty acids

[Devlin-Moyer]

(much of this was originally in #674)

The Problem:

β-oxidation of fatty acids exclusively occurs on the CoA esters of fatty acids and not free fatty acids (sources: 1, 2, 3). There are a lot of reactions in Human-GEM that appear to be β-oxidation of a fatty acid, but they involve the free fatty acid and not its corresponding acyl-CoA; I haven't figured out a way to systematically identify all such reactions, but I found a particularly large cluster of them in a set of parallel pathways (one involves the R enantiomers and one involves the S enantiomers of what would otherwise be the same compound) for catabolism of leukotriene B4:

(R) enantiomer ID	(S) enantiomer ID	Reaction	Notes
MAR03801	MAR01201	6,7-dihydro-5-oxo-12-epi-LTB4 + ATP + CoA --> 5-oxo-12(R/S)-hydroxy-eicosa-(8E,10E,14Z)-trienoyl-CoA + AMP + PPi	Acyl-CoA synthesis/activation
MAR01161	MAR01204	5-oxo-12(R/S)-hydroxy-eicosa-(8E,10E,14Z)-trienoyl-CoA + O2 --> 5-oxo-12(R/S)-hydroxy-eicosa-(2E,8E,10E,14Z)-tetraenoyl-CoA + H2O2	Acyl-CoA oxidation
MAR01163	MAR01206	5-oxo-12(R/S)-hydroxy-eicosa-(2E,8E,10E,14Z)-tetraenoyl-CoA + H2O --> 3(S),12(R/S)-dihydroxy-5-oxo-eicosa-(8E,10E,14Z)-trienoyl-CoA	2-trans-enoyl-CoA hydration
MAR01165	MAR01208	3(S),12(R/S)-dihydroxy-5-oxo-eicosa-(8E,10E,14Z)-trienoyl-CoA + NAD+ --> 3,5-dioxo-12(R/S)-hydroxy-eicosa-(8E,10E,14Z)-trienoyl-CoA + H+ + NADH	3-Hydroxyacyl-CoA oxidation
MAR01167	MAR01210	3,5-dioxo-12(R/S)-hydroxy-eicosa-(8E,10E,14Z)-trienoyl-CoA + CoA --> 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoyl-CoA + acetyl-CoA	3-Oxoacyl-CoA thiolysis
MAR01170	MAR01212	3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoyl-CoA + H2O --> 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate + CoA + H+	Acyl-CoA hydrolysis (not wrong, per se, but MAR01174 and MAR01216 should use the acyl-CoA and not the free fatty acid)
MAR01174	MAR01216	3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate + CoA --> 8(R/S)-hydroxy-hexadeca-(4E,6E,10Z)-trienoate + acetyl-CoA	Should be 3-oxoacyl-CoA thiolysis
MAR01175	MAR01217	8(R/S)-hydroxy-hexadeca-(4E,6E,10Z)-trienoate + O2 --> 8(R/S)-hydroxy-hexadeca-(2E,4E,6E,10Z)-tetraenoate + H2O2	Should be acyl-CoA oxidation
MAR01176	MAR01218	8(R/S)-hydroxy-hexadeca-(2E,4E,6E,10Z)-tetraenoate + H+ + NADPH --> 8(R/S)-hydroxy-hexadeca-(2E,6E,10Z)-trienoate + NADP+	Should be 2,4-dienoyl CoA reduction
MAR01177	MAR01219	3(S),8(R/S)-dihydroxy-(6E,10Z)-hexadecadienoate <=> 8(R/S)-hydroxy-hexadeca-(2E,6E,10Z)-trienoate + H2O	Should be 2-enoyl-CoA hydration
MAR01178	MAR01220	3(S),8(R/S)-dihydroxy-(6E,10Z)-hexadecadienoate + NAD+ <=> 3-oxo-8(R/S)-hydroxy-hexadeca-(6E,10Z)-dienoate + H+ + NADH	Should be 3-hydroxyacyl-CoA hydration
MAR01179	MAR01221	3-oxo-8(R/S)-hydroxy-hexadeca-(6E,10Z)-dienoate + CoA <=> 6(R/S)-hydroxy-tetradeca-(4E,8Z)-dienoate + acetyl-CoA	Should be 3-oxoacyl-CoA thiolysis
MAR01180	MAR01223	6(R/S)-hydroxy-tetradeca-(4E,8Z)-dienoate + O2 --> 6(R/S)-hydroxy-tetradeca-(2E,4E,8Z)-trienoate + H2O2	Should be acyl-CoA oxidation
MAR01181	MAR01222	6(R/S)-hydroxy-tetradeca-(2E,4E,8Z)-trienoate + H+ + NADPH <=> 6(R/S)-hydroxy-tetradeca-(2E,8Z)-dienoate + NADP+	Should be 2,4-dienoyl-CoA reduction
MAR01182	MAR01224	6(R/S)-hydroxy-tetradeca-(2E,8Z)-dienoate + H2O --> 3(S),6(R/S)-dihydroxy-tetradec-(8Z)-enoate	Should be 2-trans-enoyl-CoA hydration
MAR01183	MAR01225	3(S),6(R/S)-dihydroxy-tetradec-(8Z)-enoate + NAD+ --> 3-oxo-6(R/S)-hydroxy-tetradec-(8Z)-enoate + H+ + NADH	Should be 3-hydroxyacyl-CoA oxidation
MAR01184	MAR01226	3-oxo-6(R/S)-hydroxy-tetradec-(8Z)-enoate + CoA --> 4(R/S)-hydroxy-dodec-(6Z)-enoate + acetyl-CoA	Should be 3-oxoacyl-CoA thiolysis
MAR01185	MAR01227	4(R/S)-hydroxy-dodec-(6Z)-enoate + CoA --> (2E)-dodecenoyl-CoA + H+ + O2	Not real chemistry, but fixing that is complicated; see #674

With the exception of 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate, none of these free fatty acids appear to have corresponding acyl-CoA metabolites in Human-GEM already, but these are the only reactions that all of these metabolites (again except for 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate; they can also be transported to/from the cytosol) participate in, so we can remedy this situation by just editing the metabolite objects to make them represent their acyl-CoA equivalents. We'd also want to edit MAR01174 and MAR01216 to use 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoyl-CoA instead of 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate.

Editing MAR01174 and MAR01216 causes a new problem: 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate can also be produced from 6,7-dihydro-5-oxo-12-epi-LTB4 in the peroxisome and then transported into the mitochondria (which happens in real cells; source), and making MAR01174 and MAR01216 use 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoyl-CoA instead of 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate would render the peroxisomal version of that pathway a dead end. Fortunately, all we need to do to avoid this is add two new acyl-CoA activation reactions for 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate.

Proposed Changes:

• Edit MAM01206m to be:
  • name: 8(S)-hydroxy-hexadeca-(4E,6E,10Z)-trienoyl-CoA
  • formula: C37H56N7O18P3S4-
  • charge: -4
• Edit MAM01201m to be:
  • name: 8(R)-hydroxy-hexadeca-(4E,6E,10Z)-trienoyl-CoA
  • formula: C37H56N7O18P3S4-
  • charge: -4
• Edit MAM01204m to be:
  • name: 8(S)-hydroxy-hexadeca-(2E,4E,6E,10Z)-tetraenoyl-CoA
  • formula: C37H54N7O18P3S4-
  • charge: -4
• Edit MAM01199m to be:
  • name: 8(R)-hydroxy-hexadeca-(2E,4E,6E,10Z)-tetraenoyl-CoA
  • formula: C37H54N7O18P3S4-
  • charge: -4
• Edit MAM01205m to be:
  • name: 8(S)-hydroxy-hexadeca-(2E,6E,10Z)-trienoyl-CoA
  • formula: C37H56H7O18P3S4-
  • charge: -4
• Edit MAM01200m to be:
  • name: 8(R)-hydroxy-hexadeca-(2E,6E,10Z)-trienoyl-CoA
  • formula: C37H56N7O18P3S4-
  • charge: -4
• Edit MAM00695m to be:
  • name: 3(S),8(S)-dihydroxy-(6E,10Z)-hexadecadienoyl-CoA
  • formula: C37H58N7O19P3S4-
  • charge: -4
• Edit MAM00694m to be:
  • name: 3(S),8(R)-dihydroxy-(6E,10Z)-hexadecadienoyl-CoA
  • formula: C37H58N7O19P3S4-
  • charge: -4
• Edit MAM00851m to be:
  • name: 3-oxo-8(S)-hydroxy-hexadeca-(6E,10Z)-dienoyl-CoA
  • formula: C37H56N7O19P3S4-
  • charge: -4
• Edit MAM00850m to be:
  • name: 3-oxo-8(R)-hydroxy-hexadeca-(6E,10Z)-dienoyl-CoA
  • formula: C37H56N7O19P3S4-
  • charge: -4
• Edit MAM01149m to be:
  • name: 6(S)-hydroxy-tetradeca-(4E,8Z)-dienoyl-CoA
  • formula: C35H54N7O18P3S4-
  • charge: -4
• Edit MAM01146m to be:
  • name: 6(R)-hydroxy-tetradeca-(4E,8Z)-dienoyl-CoA
  • formula: C35H54N7O18P3S4-
  • charge: -4
• Edit MAM01147m to be:
  • name: 6(S)-hydroxy-tetradeca-(2E,4E,8Z)-trienoyl-CoA
  • formula: C35H52N7O18P3S4-
  • charge: -4
• Edit MAM01144m to be:
  • name: 6(R)-hydroxy-tetradeca-(2E,4E,8Z)-trienoyl-CoA
  • formula: C35H52N7O18P3S4-
  • charge: -4
• Edit MAM01148m to be:
  • name: 6(S)-hydroxy-tetradeca-(2E,8Z)-dienoyl-CoA
  • formula: C35H54N7O18P3S4-
  • charge: -4
• Edit MAM01145m to be:
  • name: 6(R)-hydroxy-tetradeca-(2E,8Z)-dienoyl-CoA
  • formula: C35H54N7O18P3S4-
  • charge: -4
• Edit MAM00693m to be:
  • name: 3(S),6(S)-dihydroxy-tetradec-(8Z)-enoyl-CoA
  • formula: C35H56N7O19P3S4-
  • charge: -4
• Edit MAM00692m to be:
  • name: 3(S),6(R)-dihydroxy-tetradec-(8Z)-enoyl-CoA
  • formula: C35H56N7O19P3S4-
  • charge: -4
• Edit MAM00848m to be:
  • name: 3-oxo-6(S)-hydroxy-tetradec-(8Z)-enoyl-CoA
  • formula: C35H54N7O19P3S4-
  • charge: -4
• Edit MAM00847m to be:
  • name: 3-oxo-6(R)-hydroxy-tetradec-(8Z)-enoyl-CoA
  • formula: C35H54N7O19P3S4-
  • charge: -4
• Edit MAM00936m to be:
  • name: 4(S)-hydroxy-dodec-(6Z)-enoyl-CoA
  • formula: C33H52N7O18P3S4-
  • charge: -4
• Edit MAM00935m to be:
  • name: 4(R)-hydroxy-dodec-(6Z)-enoyl-CoA
  • formula: C33H52N7O18P3S4-
  • charge: -4
• Replace MAM00835m with MAM00836m in MAR01174
• Replace MAM00837m with MAM00838m in MAR01216
• Create a new reaction: MAM00835m + MAM01371m + MAM01597m -> MAM00836m + MAM01334m + MAM02759m, GPR: ENSG00000068366 or ENSG00000103740 or ENSG00000123983 or ENSG00000130377 or ENSG00000140284 or ENSG00000151726 or ENSG00000164398 or ENSG00000197142 (same as MAR01201 and MAR03801)
• Create a new reaction: MAM00837m + MAM01371m + MAM01597m -> MAM00838m + MAM01334m + MAM02759m, GPR: ENSG00000068366 or ENSG00000103740 or ENSG00000123983 or ENSG00000130377 or ENSG00000140284 or ENSG00000151726 or ENSG00000164398 or ENSG00000197142 (same as MAR01201 and MAR03801)

[JonathanRob]

Good spotting, I agree that β-oxidation of FFAs should be replaced with their CoA-activated counterparts, since that is what the literature supports.

However, I'm not sure it would be ideal to edit the existing FFA metabolites (name, formula, charge) to fix the issue, since they still represent a real metabolite that maps to other identifiers in many cases. Although it is unfortunately more work, it would be cleaner to create new metabolites for these CoA-activated forms (if they do not yet exist in the model), and update the reaction formula to use these new metabolites instead.

[Devlin-Moyer]

at the moment, I don't know of any reactions that those free fatty acids could participate in, so making new metabolites for their CoA esters and replacing them in these particular reactions would leave almost all of the original free faty acids isolated, which I'm pretty sure the GitHub actions pull request checks flag as a problem. I suppose the simplest solution would be to add a CoA hydrolysis reaction for each metabolite to keep all the FFAs connected, but they'd still be dead ends. Also it'd maybe be a little weird to have acyl-CoA hydrolysis reactions for every beta-oxidation intermediate and not just the ones in between rounds of beta-oxidation, since the 2-trans-enoyl-CoA hydration, 3-hydroxyacyl-CoA oxidation, and 3-oxoacyl-CoA thiolysis reactions would all happen on the HADHA/HADHB complex (since this is all beta-oxidation of a fatty acid chain of longer than ~8 carbons in the mitochondria) in each round of beta-oxidation, which doesn't seem like it'd be conducive to an acyl-CoA thioesterase intervening in between those three steps.

[JonathanRob]

Do you mean the FFAs would be isolated in the sense that they no longer are associated with any reactions? If so, then I think it would be fine to just remove them from the model altogether. Or if they become dead-end in just one or a few reactions, it's also not a huge concern - gaps can always be filled with future work.

I think as long as it doesn't break any of the metabolic tasks, then we should be safe. Though maybe check both the Essential and Full metabolic task lists to verify that we are not accidentally disrupting any important pathways.

[Devlin-Moyer]

yea the only reactions any of these FFAs participate in (except for 3-oxo-10(R/S)-hydroxy-octadeca-(6E,8E,12Z)-trienoate, which is already accounted for in the proposed changes) are the ones in this table

[Devlin-Moyer]

Re: not wanting to edit the metabolite objects directly because they map to other identifiers: all of the metabolites that I proposed to edit are only associated with EHMN, Recon3D, and HMR2 metabolite IDs in model/metabolites.tsv. I looked up the reactions that a few of the Recon3D equivalents participate in in Recon3D, and they're also not-quite-right fatty acid oxidation reactions using free fatty acids instead of acyl-CoAs. If they were always "supposed" to represent their acyl-CoA equivalents, is it necessary to make entirely new metabolite objects?