Ubiquinol biosynthesis (R-HSA-2142789)

[ukemi]

• Make mouse models-
• http://noctua.geneontology.org/editor/graph/gomodel:67369e7600006071?

[ukemi]

R-HSA-9856502 and R-HSA-9856510 are given the molecular functions of lipid transfer activity (GO:0120013). But according to ChEBI, ubiquinol 10 (CHEBI:64183) is not a lipid.

[ukemi]

Since mice can make ubiquinol 10, but mostly make ubiquinol 9, should I:

1. Switch all the deca molecules to the corresponding nona molecules and the corresponding enzymes for which we have specificity?
2. Keep the deca pathway and create a separate nona pathway?
3. Switch all of the deca molecules to a more generic parent that is chain length agnostic, and change the enzymes?

@sjm41 @rozaru @deustp01 @vanaukenk what do you think?

[sjm41]

I'd say option 3.

Anyway, we are eliminating all of the chain-length specific MF terms relevant to the pathway, so there should only be "polyprenyl" type terms left, though those changes may not all be in Noctua yet.

Though you can of course still choose to specify a chain length as the input/output chemcial if you really want to. (I think some species are meant to only make a single chain length, or so the literature says....)

FYI, here's Helen's draft model of the fly pathway, which is still pending some of the newer MF terms to come through the system.
http://noctua.geneontology.org/workbench/noctua-visual-pathway-editor/?model_id=gomodel%3A67369e7600004023

[ukemi]

If option 3 is the way things a GOing, then I will change my model when the new terms are ready. Are the specific MFs going to be obsoleted? If so I will make my model 'development' and spare Li the pain of changing them. @deustp01 this means we should probably change the Reactome mappings where appropriate.

[sjm41]

Are the specific MFs going to be obsoleted?
Our recent review only found a couple of extant MF/BP for ubiquinol biosynthesis that were chain-length specific, and yes, those have been changed/obsoleted in preference of the length/species-agnostic versions.

Can you list the length-specific MFs currently used in your model?

[ukemi]

[GO:0097269] all-trans-decaprenyl-diphosphate synthase activity
[GO:0036169] 3-methoxy-4-hydroxy-5-decaprenylbenzoic acid decarboxylase activity

replaced with generic parents in the model.

[sjm41]

[GO:0036169] 3-methoxy-4-hydroxy-5-decaprenylbenzoic acid decarboxylase activity

That term has only been used in a single annotation (the TAS from Reactome) and it has no xrefs to EC/RHEA.
The good news is that we made a new term earlier this week that covers this COQ4 reaction (length-agnostic), based on new EC/RHEA entries- see #29459:
GO:0120539
4-hydroxy-3-methoxy-5-polyprenylbenzoate decarboxylase activity

I guess this won't be available in Noctua just yet, but I'd suggest you use it when it is!

I'll make another ticket to merge GO:0036169 into GO:0120539.

[sjm41]

[GO:0097269] all-trans-decaprenyl-diphosphate synthase activity

This is one of the terms that can be used for the 'initial' COQ1 (PDSS) reaction. It's this enzyme that determines the chain length (as we understand the process), so we're keeping all the GO-MF terms for this initial step, but not for the subsequent steps (since those enzymes are not strictly chain length specific - they just work with whatever COQ1/PDSS produces.

However, for flies and mice (at least), where the chain length appears to be variable, you might still want to use a the more generic parent - I recently made such a term, so I'd recommend you use that:
id: GO:0120531
name: prenyl diphosphate synthase activity

[ukemi]

Thanks @sjm41! I will update my model when the terms are available. You are correct, in mice the chain length is predominantly 9, but 10 is also made as a minor species.

[ukemi]

Or, maybe merge 'all-trans-nonaprenyl-diphosphate synthase (geranyl-diphosphate specific) activity' (GO:0050347) and 'all-trans-nonaprenyl-diphosphate synthase (geranylgeranyl-diphosphate specific) activity' (GO:0052924) into a new generic parent, 'all-trans-nonaprenyl-diphosphate synthase activity'. But then we need a generic input and a new Rhea reaction too, I think??????. Reactome uses ChEBI[175763] and ChEBI[128769] as the inputs and I have retained those, but I replaced the chemicals with the nona forms downstream of this. https://reactome.org/PathwayBrowser/#/R-HSA-2162253

[deustp01]

R-HSA-9856502 and R-HSA-9856510 are given the molecular functions of lipid transfer activity (GO:0120013). But according to ChEBI, ubiquinol 10 (CHEBI:64183) is not a lipid.

But according to these publications and a Google AI response, the ChEBI ontology is incomplete.
From https://www.lipotype.com/lipidomics-services/prenol-lipid-analysis/quinone-analysis/ubiquinone-ubiquinol-analysis/

Ubiquinones/ubiquinols (coenzyme Q, mitoquinones, CoQ, or UQ) belong to the group of quinones within the prenol lipids category. Their structure consists of a single- or double-ring structure linked to a chain made of five-carbon isoprene units, the building blocks of prenol lipids.

And from https://www.sciencedirect.com/topics/medicine-and-dentistry/ubiquinone#:~:text=Ubiquinone%20(Coenzyme%20Q%20or%20CoQ,usually%206%E2%80%9310%20units).

Ubiquinone (UQ) ... is a lipid consisting of a quinone head group and a polyprenyl tail varying in length depending on the organism.

The ChEBI classification takes account only of the quinone head group and ignores the polyprenyl tail. If ChEBI treats nucleotides as kinds of sugars, it can treat ubiquinone as kinds of lipids.

[deustp01]

MF cleanup

@ukemi said, this means we should probably change the Reactome mappings where appropriate.

Specifically, per @sjm41

[GO:0036169] 3-methoxy-4-hydroxy-5-decaprenylbenzoic acid decarboxylase activity
That term has only been used in a single annotation (the TAS from Reactome) and it has no xrefs to EC/RHEA.
The good news is that we made a new term earlier this week that covers this COQ4 reaction (length-agnostic), based on new EC/RHEA entries- see #29459: GO:0120539

So if I understand right, replacing GO:0036169 in reaction R-HSA-2162195 "COQ4 decarboxylates MHDB" with the new term GO:0120539 would solve the problem discussed here. This does not lose any information because the specific inputs and outputs are annotated separately.

Done - if I've missed something here or additional changes are needed, let me know. The change will become publicly visible in our March 2025 release