GPRs for Mitochondrial Acyl-CoA Dehydrogenase Reactions

[Devlin-Moyer]

trafficstars

Current behavior:

As briefly mentioned in #607, some mitochondrial reactions in Human-GEM are associated with genes that catalyze analogous reactions in peroxisomes and not mitochondria. I found a total of 128 reactions that involve at least one metabolite in [m] and at least one enzyme known to be involved in fatty-acid oxidation in peroxisomes and not mitochondria. Trying to address all of them in a single issue proved to be unwieldy, so I've moved most into separate issues, and have edited this one so it only addresses the acyl-CoA dehydrogenation reactions:

ID	Reaction	Chain Length	Unsaturated?	Current GPR	Proposed New GPR
MAR01253	20-CoA-20-oxo-LTB4 + O2 --> (18E)-20-oxo-20-CoA-LTB4 + H2O2	20	yes	ACOX3 or ACOX1	ACAD9 or ACADL or ACADVL (ENSG00000072778 or ENSG00000115361 or ENSG00000177646)
MAR01270	10,11-dihydro-LTB4-CoA + O2 --> 5(S),12(R)-dihydroxy-eicosa-2,8-trans-6,14-cis-tetraenoyl-CoA + H2O2	20	yes	ACOX3 or ACOX1	ACAD9 or ACADL or ACADVL (ENSG00000072778 or ENSG00000115361 or ENSG00000177646)
MAR03997	O2 + Cis-6-Dodecenoyl Coenzyme A --> H2O2 + trans,cis-lauro-2,6-dienoyl-CoA	12	yes	ACOX1	ACADL or ACADVL (ENSG00000072778 or ENSG00000115361)
MAR03291	FADH2 + Trans,Cis-Dodeca-2,5-Dienoyl Coenzyme A <=> (5Z)-dodecenoyl-CoA + FAD	12	yes	ACADL or ACADM or ACOX1 or ACADSB	ACADL (ENSG00000115361)
MAR03163	butyryl-CoA + FAD --> crotonyl-CoA + FADH2	4	no	ACAD10 or ACADM or ACADS or ACAD8 or ACAD9 or ACADSB or ACAD11	ACADS or ACADSB or ACAD8 (ENSG00000122971 or ENSG00000151498 or ENSG00000196177)
MAR03212	FAD + propanoyl-CoA --> acrylyl-CoA + FADH2	3	no	ACAD10 or ACADM or ACADS or ACAD8 or ACAD9 or ACADSB or ACAD11	ACADS or ACADSB or ACAD8 (ENSG00000122971 or ENSG00000151498 or ENSG00000196177)

Expected behavior:

ACOX1 and ACOX3 catalyze the oxidation of acyl-CoAs into 2-trans-enoyl-CoAs in peroxisomes, so neither of them should be associated with any of these reactions. The enzymes responsible for catalyzing the same reaction in mitochondria are ACADVL, ACADL, ACADM, ACADS, ACADSB, ACAD8, ACAD9, and ACAD11 (see this paper for a general overview of the differences between mitochondrial and peroxisomal beta-oxidation reactions (specifically figure 1)). The ACADs all have different preferences/affinities for acyl-CoAs with different chain lengths, which is why I included a column for the chain length in the above table, and I've tried to ensure my proposed new GPRs reflect these preferences:

Enzyme	Chain Lengths	Can Handle Unsaturation?	Sources
ACADVL	12-24	yes	figure 4E-H, figure 2 and table 3
ACADL	6-24	yes	figure 4E-H, figure 3
ACADM	6-16	no	figure 4E-H, figure 4
ACADS	4	no	figure 4E-H
ACADSB	<=6	no	table 2
ACAD8	3-4	no	figure 3
ACAD9	9-11, 14-22	yes	figure 4
ACAD11	20-24	unknown	figure 4E-H

Proposed Changes:

• [x] Set GPRs of MAR01253 and MAR01270 to ENSG00000072778 or ENSG00000115361 or ENSG00000177646
• [x] Set the GPR of MAR03997 to ENSG00000072778 or ENSG00000115361
• [x] Set the GPR of MAR03291 to ENSG00000115361
• [x] Set GPRs of MAR03163 and MAR03212 to ENSG00000122971 or ENSG00000151498 or ENSG00000196177

[feiranl]

Nice Result! Will check this soon. May I know where those protein localizations from? We updated genes.tsv with the protein localization from HPA (#437) and DeepLoc2(#440) prediction. Is there any conflict localization documentation with what you proposed in this issue?

[haowang-bioinfo]

As briefly mentioned in https://github.com/SysBioChalmers/Human-GEM/issues/607, some mitochondrial reactions in Human-GEM are associated with genes that catalyze analogous reactions in peroxisomes and not mitochondria. I found a total of 128 reactions that involve at least one metabolite in [m] and at least one enzyme known to be involved in fatty-acid oxidation in peroxisomes and not mitochondria:

good work

There are some other issues with some of these reactions, but I'll address those in a different issue so that this one can just be about changing which genes are associated with each.

please do so - it's alway nice to have issues with specific request, while overview discussion is also good

[Devlin-Moyer]

@feiranl I got the localizations from the links in the second table (i.e. Uniprot or various papers when Uniprot did not have a localization or only had a localization "by similarity" or something else that made me suspicious); I did not check genes.tsv/HPA/DeepLoc2, but I can do that

[Devlin-Moyer]

I've edited my recommended new GPRs to align with the localizations given in genes.tsv from SwissProt/HPA/DeepLoc2 in all cases except:

• ACSF2: Uniprot says it is only localized to mitochondria, but that is a "manual assertion by curator" with no publications attached; HPA says it's cytosolic and nuclear, but is "uncertain" about that. For now, I'm gonna say it's okay to associate ACSF2 with both [c] and [m] reactions, and if anyone ever figures out its precise subcellular localization, that can be revisited
• SLC27A3: HPA seems quite confident that it's only localized to the E.R., but Uniprot cites this paper that did multiple experiments that pretty clearly showed that SLC27A3 was definitely in mitochondria and probably not in the E.R., so I think it should stay on my list of mitochondrial acyl-CoA synthetases.
• SLC25A5: Uniprot says it's localized to the E.R. (which is what's in genes.tsv) and the cell membrane, and most of the papers I can find about it (e.g. the one Uniprot cites) talk about its role as a transporter, so I figured that meant it might be capable of catalyzing CoA esterification of fatty acids on both sides of the E.R. and cell membranes, which is why I have it listed as a "Cytosolic" acyl-CoA synthetase
• SLC27A6: HPA says it's localized to "nuclear bodies", but Uniprot cites this paper that pretty clearly demonstrated that SLC27A6 is in the plasma membranes of cardiac cells and capable of transporting fatty acids across it, which is why I have it listed as a "Cytosolic" acyl-CoA synthetase