definition of GO:0015091

[gocentral]

The definition of "ferric iron transmembrane transporter activity" (GO:0015091) is "Catalysis of the transfer of ferric iron (Fe(III) or Fe3+) ions from one side of a membrane to the other." I don't think the use of "catalysis" in the definition is appropriate. "catalysis" involves modification of a chemical reaction, and transport of ions is not a chemical reaction. I think the definition would be more accurate as this: "The transfer of ferric iron (Fe(III) or Fe3+) ions from one side of a membrane to the other." I also suggest parallel changes in other transporter definitions.

Reported by: slaulederkind

Original Ticket: geneontology/ontology-requests/11546

[gocentral]

Amplifying on the "other transporter definition" part of Stans's request, there is a disconnect in the molecular_function parentage. Tracing up the lineage, parents and grandparents are likewise defined as "catalysis of ...", all the way up the great-great-great-great grandparent GO:0015075 ion transmembrane transporter activity, "Catalysis of the transfer of an ion from one side of the membrane to the other." Its parent, GO:0022891 substrate-specific transmembrane transporter activity, however, is defined as "Enables the transfer of a specific substance or group of related substances from one side of a membrane to the other."

Original comment by: deustp01

[gocentral]

• assigned_to: Stan Laulederkind --> David Hill

Original comment by: paolaroncaglia

[gocentral]

Assigning to David Hill as he was on duty last week. Thanks David. :-)

Original comment by: paolaroncaglia

[gocentral]

According to the wiki page for the transport project, these were the accepted standard definitions. There was a note as to whether the active transport terms should have the word catalysis in their definitions and the passive transport terms should have the word enables.

I am hesitant to change definitions that were agreed upon at a meeting with experts. I will bring this up on an editors' call.

-D

Original comment by: ukemi

[gocentral]

Hi David,

I was in the "transport" group. Most of this was done by Jen from textbook. I don't think we had much input from experts so it could be the case that some of these definitions do not match the community consensus. I would not like to say that what we have already is completely correct.... Catalysis here does seem odd to me (because logically, based on this, we would need to house transporters under catalytic activity and metabolic process).

It would be useful if we could trace this change to a specific meeting....I know that the meetings were documented on awiki page....do we know where this is? I can't find them here http://wiki.geneontology.org/ ...is the old wiki still active/archived? I don't remember the URL...

val

Original comment by: ValWood

[gocentral]

In Wikipedia "catalysis" (http://en.wikipedia.org/wiki/Catalysis) "... is the increase in the rate of a chemical reaction due to the participation of an additional substance called a catalyst" as a generality (parent). "Enzymes" http://en.wikipedia.org/wiki/Enzyme "... accelerate, or catalyze, chemical reactions" and a membrane transport protein (http://en.wikipedia.org/wiki/Membrane_transport_protein) "... (or simply transporter) is a membrane protein involved in the movement of ions, small molecules, or macromolecules, such as another protein, across a biological membrane". These definitions make enzyme activity and transport sound like specific forms of catalysis (children). I have no idea what an expert enzymologist would say about this, but to the extent that Wikipedia represents an expert consensus opinion, Val's "logic" looks plausible. Right now, although molecular_function has transporter activity and catalytic activity as children, all the children of catalytic activity describe activities of enzymes so in effect the functions of enzymes and transmembrane transporters fall into sibling classes.

Original comment by: deustp01

[gocentral]

I am happy with switching these to enables, but will leave them open for a bit until the discussion dies down.

Original comment by: ukemi

[gocentral]

It might be good to get an expert opinion here. As the Oliver lab use both transporters and enzymes in metabolic models, I will ask their opinion on this thread too, to see what their expectation would be.

In fact I think Steve was one of the experts we recruited for the transport overhaul if I recall correctly.

Val

Original comment by: ValWood

[gocentral]

Response from Duygu in Steve's lab: Some useful insights. Val

It is a very interesting thread, especially for me, because I have spent almost half of last year to improve the iron transport and exchange mechanisms in the Sc metabolic model :) I have added Ayca to this conversation so that she can correct me if I say something implausible. Her PhD thesis was on the interface between the iron and copper mechanisms in Sc. The exact GO term does not exist for Sc but instead, a very similar (slightly more general) term does: GO:0005381 iron ion transmembrane transporter activity. In Sc, FTH1, FTR1, MRS3, MRS4, FET3 and FET4 are associated with the term.

I am not a biochemist, nor an expert on catalysis but if I were given the chance, I would go for keeping the term. There are interesting aspects to it: First of all, by definition, Ftr1p is a permease (which kind of alerts me to the presence of an enzyme), Fet3p is a ferro-O2-oxidoreductase. They form a complex (catalyzed by the incorporation of copper ions) for the transport reaction to take place. And this complex, while transporting iron across the membrane, also facilitates the oxidoreductive reaction between the ferric and the ferrous forms.

On the other hand, Fth1p, Mrs3p, Mrs4p and Fet4p (other genes annotated with the same term) are transporters with no known catalytic actions (they just facilitate the transport of the molecule).

As for other similar catalytic activity happening during transport, there is experimental evidence in the literature for a similar oxidoreductive transport of copper (involving cuprous and cupric forms) (by Fre2p in Sc). And also mitochondrial oxidoreductive transport of iron (through Yfh1p again in Sc).

So my instinct would be to keep the term but to revise the annotated genes.

I have not studied other ion transport systems, so I cannot say anything on their behalf.

I hope I am making some sense, every time I go through through this mechanism, it appears to be more complicated than before:)

All the best, Duygu

Original comment by: ValWood

[gocentral]

My follow up:

Fantastic, that is really helpful. In, fact we have a process term for these permeases:

GO:0033215 iron assimilation by reduction and transport A process in which iron is solubilized by reduction from Fe3+ to Fe2+ via a cell surface reductase and subsequent transport of the iron across the membrane by iron uptake proteins.

The crux of the issue seems to be that in GO is that this activity is considered multistep, and so at the MF level, historically we separated individual steps into their component molecular functions. Accordingly this process has two molecular function "has_parts" : Has part GO:0034755 iron ion transmembrane transport Has part GO:0000293 ferric-chelate reductase activity

I'm not convinced that this separation is completely appropriate if the reactions are 'coupled' so we probably need to revisit this. Is GO:0033215 really separable into two individual steps?

Is it OK if I post your response into the SF ticket? I think it points us in the right direction. It would be good to use the iron transporters as a focus for a clean up of the ontology and the annotation, and then if appropriate we could roll the changes out over the transporter node.

We will probably have further questions too!

Original comment by: ValWood

[gocentral]

Then it gets more complicated:

I now see that these are separate protein components a reductase and a permease, so probably the reduction and transport steps really are separable activities here ?

Either way, here the "ferric-chelate reductase activity" is an integral part of the iron transmembrane transport process so I would probably recommend:

GO:0034755 iron ion transmembrane transport (BP) -- (part_of) GO:1903874 ferrous iron transmembrane transport (BP) --- (part_of) GO:0033215 ferrous transmembrane transport by reduction and transport (BP) (the existing assimilation term) ---- (has_part) GO:0000293 ferric-chelate reductase activity (MF) ---- (has_part) GO:0015093 ferrous iron transmembrane transporter activity (MF)

i.e rename, move and redefine GO:0033215 assimilation term which does not currently have transport parent. Replace the has_part GO:0034755 by the more specific MF term. This would propagate annotations correctly from GO:0033215 to the iron iron transmembrane transport terms.

Original comment by: ValWood

[gocentral]

More from Duygu

Dear Val,

As I have said before, I would not be the person to judge the single-step or multi-step behaviour of this phenomenon. However, from what I have read, I would assume that the formation of a single complex is pretty much required. Below is a script from the summary paragraph for FTR1 in SGD:

"For proper cell-surface targeting, the Fet3p-Ftr1p complex must be assembled in the endoplasmic reticulum 9. Without Fet3p, Ftr1p is retained in the ER 1. Without Ftr1p, Fet3p proceeds as far as the Golgi apparatus, but is misglycosylated and recycled back to the ER 10."

Even if it were not the case for these two enzymes, Yfh1 does what the Ftr1-Fet3 complex does on the cell periphery across the mitochondrial membrane (and as far as known, there is a single enzyme responsible for both oxidoreduction and transport).

Of course I would not mind you sharing my e-mails, as long as people are warned ahead that I am not an expert and everything I say should be approached with caution.

All the best, Duygu

Original comment by: ValWood

[gocentral]

So, I can see why we have the current situation but it will require some discussion to figure out what the current policy is first.

Also, I think we might do different things in different contexts. For example, some enzyme regulators (like a cyclin CDK partner) are required for enzyme activity, but we don't annotate necessarily these to the enzyme activity per. se. This might be a similar situation. Lots of things are required for something else to occur, but we need to to figure out when it is appropriate to lump and when to split, and to do this consistently.

.....Of course the problem would also go away if we could assign GO terms easily to a "complex" object rather than only to individual gene products.

As far as I can see the outcome hinges on whether the catalysis and transport steps are always separable. I don't know enough about transport mechanisms, or the GO policy to figure this out

Val

Original comment by: ValWood

[gocentral]

Hi Everyone,

Reading this thread I am tempted to define the transporters with the enabled phrasing rather than the catalysis phrasing. My rationale is that these terms were not meant to represent an activity of any single gene product, but rather to represent the transporter reaction. Keeping that in mind, it seems that the transporter reaction in its most generic form is not a catalysis, but that some gene products might couple the transporter function with a catalytic event. In that case, I think it would be appropriate to either create a new term that describes both (inseparable) activities or to create multiple annotations. Is everyone ok with this??

-D

Original comment by: ukemi

[gocentral]

I agree. Which is basically Stan's original request... sorry for the divergence but it is good to revisit what we did and why.

I prefer to make the co-annotation to the catalysis and transport step if both are valid for a gene product. This way transport and catalysis will/should? be disjoint in the ontology (Is that correct?), and when the catalysis and transport steps are separate or coupled it will be much clearer (I will be able to evaluate these much more clearly in the QC matrix with this distinction).

Original comment by: ValWood

[gocentral]

Only just seen this:

Val writes: ".....Of course the problem would also go away if we could assign GO terms easily to a "complex" object rather than only to individual gene products. "

You can do this now! IntAct complex ACs are now allowed as annotation objects in P2GO so if you ever need a complex as an annotation object, send me the details and I curate the complex (if it's not already in the CP). I just need the PMID that shows physical interaction evidence for the complex (or some other clear indication we can infer the existence of the complex from such as pharmacological evidence for membrane receptors...). A proposed Def and any GO annotations and evidences you have for the complex function would be helpful as well. I have a JIRA tracker for requests but I have to ask EBI systems admin to add new users on an individual basis so can't simply give you the URL as it's password protected.

In the current example the complex would have the transporter and catalysis annotation extensions linking it to both classes. Of course there are many transporters that don't catalyse the molecule but 'simply' facilitate its transport through the membrane (but in an 'active' way as opposed to a pore - not opening that can of worms here!). On the other hand, a lot of transporters have an ATPase subunit that enables the transporter function... (e.g. ABC transporters, I've done a fair few of them for ecoli)

Birgit

Original comment by: bmeldal

[gocentral]

Hi Birgit,

Our problems are mainly that even once complex IDs are avaialable we will need changes to our curation database, curation tool and web pages. That said we recently mad a move to obtain complex IDs from PRO for all of the complexes that we use. Are these 2 efforts (PRO and Intact) aligned in any way? It seems a shame to have 2 sets of accession numbers and the duplication of effort....

Val

Original comment by: ValWood

[gocentral]

PRO and Intact are currently trying to sort out id space. IntAct has a nice interface to make the complexes but does not make PRO ids. There are some issues with mapping/translating pro to Intact and back which I am told is being worked on. MGI is in the same boat that we currently cannot directly import/display any sort of complex ids as they would map to multiple genes. Any GO annotations that we have made to PRO complexes resides at PRO at the moment.

Original comment by: hdrabkin

[gocentral]

Yes, that's right. If there is a PRO ID for a complex we add it as a xref but as our data schemas and file formats are different a direct import/export is not possible. We, at Intact, work off an obo file that Darren sent. I did a quick search in our latest release and we only seem to have 17 PRO complexes. Seems like the overlap is not great :(

Seems like the changes to P2GO don't help everyone which is a shame.

We are happy to talk more details to get curation efforts aligned. In any case, our Editor is available to all (you just need a username and password) so we can train anyone in using it and building complexes themselves.

Birgit

Original comment by: bmeldal

[gocentral]

Hi Val et al,

Looking at this again while starting the edits. What about 'active transmembrane transporter activity'? Seems to me that these should be catalysis.

-D

Original comment by: ukemi

[gocentral]

That does sound like it should be catalysis! So does that mean that "active" are catalytic, and "passive" are not.

Original comment by: ValWood

[gocentral]

That is the split that I propose. The active transporters will retain the catalysis of... defs and the others will be switched to enable.

-D

Original comment by: ukemi

[gocentral]

Sounds good. Thank you, David.

Original comment by: slaulederkind

[gocentral]

• status: open --> closed-fixed

Original comment by: ukemi

[gocentral]

Wow. That was a lot of edits. Let me know if you find any stragglers.

-D

Original comment by: ukemi