FAD/Ubiquinone Duplicate Reactions

[Devlin-Moyer]

The Problem:

This is follow-up of what discussed in #599

These pairs of reactions all represent reactions that are catalyzed by enzymes that use FAD as a cofactor, and the electrons harvested from their substrates eventually make their way from FAD to ubiquinone, but Human-GEM currently has one version that involves FAD and another that directly reduces ubiquinone with no explicit acknowledgement of the role of FAD:

substrates	ubiquinone-dependent ID	FAD-dependent ID
palmitoyl-CoA, (2E)-hexadecenoyl-CoA	MAR02035	MAR03121
propanoyl-CoA, acrylyl-CoA	MAR03211	MAR03212
isobutyryl-CoA, methacrylyl-CoA	MAR03751	MAR03752
2-methylbutyryl-CoA, tiglyl-CoA	MAR03783	MAR03874
glutaryl-CoA, crotonoyl-CoA	MAR04242	MAR04243
3-methylcrotonoyl-CoA, isovaleryl-CoA	MAR03769	MAR03770
butyryl-CoA, crotonoyl-CoA	MAR02366	MAR03136
lauroyl-CoA, (2E)-dodecenoyl-CoA	MAR02367	MAR03135
myristoyl-CoA, (2E)-tetradecenoyl-CoA	MAR02370	MAR03128
hexanoyl-CoA, (2E)-hexenoyl-CoA	MAR02372	MAR03156
octanoyl-CoA, (2E)-octanoyl-CoA	MAR02369	MAR03149
decanoyl-CoA, (2E)-decenoyl-CoA	MAR02373	MAR03142
1-pyrroline-5-carboxylate, proline	MAR03838	MAR08611
fumarate, succinate	MAR04652	MAR08743
DHAP, sn-glycerol-3-phosphate	MAR00483	MAR00482, MAR00449, MAR01169 (see #599)

Aside from the last few reactions, if you look at any of the genes associated with any of these reactions, the overwhelming majority of other reactions associated with thse genes only have FAD-dependent versions (e.g. out of the 90 reactions ACADVL is associated with, MAR02035 is the only one that uses ubiquinone instead of FAD), so for consistency, it would make sense to just drop the ubiquinone-dependent member of each of these pairs.

Speaking of the genes associated with these reaction, some of them (e.g. ACOX1 and ACOX3) are peroxisomal and not mitochondrial, but they're also associated with a bunch of other mitochondrial reactions, so I'm making a separate issue about that.

I put this list together by manually looking into every reaction that involves ubiquinone; it probably could've been automated, but now that I've already done it manually, I don't see a reason to figure out how to automate it, since it's not clear if that automation would be applicable to any other situations that this issue wouldn't resolve.

Update: it turns out that MAR03769 was the only reaction that ENSG00000128928 (Isovaleryl-CoA Dehydrogenase; IVD) was associated with, but it is clearly also appropriate to associate it with MAR03770 (see Uniprot), so in order to avoid making IVD no longer associated with any reactions by removing MAR03769, it should be added to the GPR of MAR03770.

Update: I found another pair like this (MAR02369 and MAR03149) and added it to the table above and list below

Suggested Changes:

• Remove MAR02035 for being a duplicate of MAR03121
• Remove MAR03211 for being a duplicate of MAR03212
• Remove MAR03751 for being a duplicate of MAR03752
• Remove MAR03783 for being a duplicate of MAR03784
• Remove MAR04242 for being a duplicate of MAR04243
• Remove MAR03769 for being a duplicate of MAR03770
• Remove MAR03838 for being a duplicate of MAR08611
• Remove MAR02366 for being a duplicate of MAR03136
• Remove MAR02367 for being a duplicate of MAR03135
• Remove MAR02370 for being a duplicate of MAR03128
• Remove MAR02372 for being a duplicate of MAR03156
• Remove MAR02369 for being a duplicate of MAR03149
• Remove MAR02373 for being a duplicate of MAR03142
• Remove MAR00483 for being a duplicate of MAR00482/MAR00449/MAR01169 (but this may already get taken care of in Proposed fixes to DHAP <=> sn-glycerol-3-phosphate reactions #599)
• Remove MAR08743 for being a duplicate of MAR04652 (only case where we're removing the FAD-dependent version and not the ubiquinone-dependent version, since this is the only reaction where the FAD is covalently bound to the enzyme catalyzing it and not in theory capable of dissociating from the enzyme between redox of the substrate and redox of ubiquinone)
• Add ENSG00000128928 to GPR of MAR03770 so it is still associated with at least one reaction

[JonathanRob]

Really nice work @Devlin-Moyer.

What do you think about the fact that by removing the direct ubiquinone-reducing reaction, the mechanism can instead use another electron acceptor to oxidize FADH2? I assume the direct ubiquinone-reducing versions were formulated to explicitly assign ubiquinone as the electron acceptor (likely to represent enzymes acting in the mitochondrial membrane), but the separation into 2 steps allows FADH2 oxidation by other means, such as:

Reaction ID	Equation
MAR08743	FADH2 [m] + fumarate [m] ⇔ FAD [m] + succinate [m]
MAR03784	2-methylbutyryl-CoA [m] + FAD [m] ⇔ FADH2 [m] + tiglyl-CoA [m]
MAR03770	3-methylcrotonyl-CoA [m] + FADH2 [m] ⇔ FAD [m] + isovaleryl-CoA [m]
MAR03752	FADH2 [m] + methacrylyl-CoA [m] ⇔ FAD [m] + isobutyryl-CoA [m]

I haven't investigated in detail, so maybe these "alternatives" are actually dead-ends that don't work, or end up using ubiquinone anyway?

[Devlin-Moyer]

As I just mentioned over in #609, allowing FAD(H₂) to move freely between any human FAD-dependent enzymes may be an accurate representation of reality: #609 (comment)

Although I now notice that SDHA is listed as one of the only the 6 human flavoenzymes that covalently bind their FAD, and SDHA catalyzes MAR04652 and MAR08743, so maybe in that case we should keep the ubiquinone-dependent version (MAR04652) and remove the FAD-dependent version to reflect the fact that the FAD bound to SDHA is (probably) not in equilibrium with free FAD, at least not to the extent that the FAD associated with the rest of the human flavoenzymes potentially is

[Devlin-Moyer]

Also the other three examples you listed are all catalyzed by enzymes that transfer electrons from their FAD cofactors to ETF, as discussed in #608, and since ETF is free to diffuse throughout the mitochondrial matrix and is clearly capable of reversible redox of its FAD cofactor, allowing transfer of FAD(H₂) between reactions associated with those enzymes in particular may not be particularly unrealistic

[JonathanRob]

Although I now notice that SDHA is listed as one of the only the 6 human flavoenzymes that covalently bind their FAD, and SDHA catalyzes MAR04652 and MAR08743, so maybe in that case we should keep the ubiquinone-dependent version (MAR04652) and remove the FAD-dependent version to reflect the fact that the FAD bound to SDHA is (probably) not in equilibrium with free FAD, at least not to the extent that the FAD associated with the rest of the human flavoenzymes potentially is

Sounds like a reasonable approach.

Also the other three examples you listed are all catalyzed by enzymes that transfer electrons from their FAD cofactors to ETF, as discussed in #608, and since ETF is free to diffuse throughout the mitochondrial matrix and is clearly capable of reversible redox of its FAD cofactor, allowing transfer of FAD(H2) between reactions associated with those enzymes in particular may not be particularly unrealistic

Agreed, thanks again for the helpful responses.

[haowang-bioinfo]

For these pairs of reactions, I assume the GPRs have been checked and confirmed the association to FAD, instead of ubiquinone @Devlin-Moyer?

Will there be other suggested changes to their GPRs than removal of peroxisomal ones (e.g. ACOX1 and ACOX3)?

[Devlin-Moyer]

Yea if you look up any of the genes currently mentioned in these GPRs on Uniprot, they all say they have an FAD cofactor.

It turns out ACOX1 and ACOX3 are also associated with a number of other mitochondrial reactions, so I'm working on a separate issue to identify all of those reactions and remove ACOX1 and ACOX3 (it's complicated by the fact that ACOX1 and ACOX3 are the only genes associated with some mitochondrial reactions, so I'm trying to figure out if those reactions just don't happen in mitochondria or if some other enzyme actually localized to mitochondria catalyzes them)

[haowang-bioinfo]

It turns out ACOX1 and ACOX3 are also associated with a number of other mitochondrial reactions, so I'm working on a separate issue to identify all of those reactions and remove ACOX1 and ACOX3

very good idea