Review annotations using extension 'activated_by'

[pgaudet]

Hello,

As discussed on the last Annotation call - I am creating a ticket to review annotations using the extension 'activated_by' We'd like to remove the relation 'activated_by'. Please review your annotations here:

https://docs.google.com/spreadsheets/d/1bUQ7uSHaLbr8uvYu04Jbmr76bn-ja01hJxbcGnx-U6I/edit#gid=0

Assigned by	Number of annotations still to fix
AgBase	1
ARUK-UCL	25
BHF-UCL	30
CAFA	5
dictyBase	23
MGI	1
ParkinsonsUK-UCL	24
PomBase	58
SGD	46
UniProt	9
WB	1

Thanks, Pascale

[RLovering]

Sorry I arrived late at this meeting http://wiki.geneontology.org/index.php/Annotation_Conf._Call_2019-11-26 please confirm whether all AE activated by data should not be captured or whether other options are going to be considered. for eg, if we created a binding annotations would this then include part_of BP extension? Thanks Ruth

[ValWood]

https://github.com/pombase/curation/issues/2484 @Antonialock has a ticket for this. Should be done shortly.

[pfey03]

I said that I have some qualms about changing or deleting some of the activated_by. For example the p2xA-E Ca Channels in Dicty have been shown to have this activity only when activated by ATP. PMID:22736763

p2xA has been shown to be located in the contractile vacuole (CV) membrane and oriented such that the putative ATP binding site is in the Vacuole lumen. In the above paper they show that ATP is translocated into the CV and only when this translocation is happening, there is Ca-channel activity by p2x channels. In a single p2xA KO the ATP activated Ca-Channel activity is strongly reduced in the CV, when all 5 are knocked out it's abolished (Fig 6A/B).

Thus ATP binding is not explicitly shown, but the binding site is where it should be in the CV lumen and without ATP there is no channel activity.

I don't like deleting the fact that the ATP activation is necessary, or changing with described methods also doesn't work as ATP binding hasn't been directly shown.

One solution might be:

There is GO:0005260 intracellularly ATP-gated chloride channel activity Although this is for a specific enzyme, I wonder if a new GO term could be an option here: " intracellularly ATP-gated calcium channel activity "

@pgaudet and @vanaukenk what would you think about this new term? Would like to get your input before opening a ticket for new term. Thanks!

[ValWood]

I was also wondering if there are alternative ways to capture some of this information. here we have a single-stranded DNA binding that requires ATP -> request ATP-dependent single-stranded DNA binding

[rjdodson]

I have found an example in dicty in which the enzymatic activity of a phosphodiesterase requires bivalent cations for activity. The source publication is PMID:11171061 and the authors present evidence in Fig.7 showing minimal enzyme activity in the absence of Mn and Mg, but significantly enhanced activity in the presence of these cations. The protein sequence of the enzyme contains two putative binding sites for bivalent metal ions, but the authors do not demonstrate that the enzyme binds the metal cations. For the enzyme activity I have currently curated in annotation extension with relation activated_by: Mn(2+) and Mg(2+). If I remove activated_by extension, how then to curate so that this result is retained?

[srengel]

SGD annotations have been reviewed. all activated_by's removed.

[pfey03]

I have one Rab GAP 'GTPase inhibitor activity' activated by Ca 2+. Here I have also a calcium binding annotation and added the function in the extension by 'directly_positively_regulates'

However, I still left the primary function annotation as well in this linear annotation. I guess in GO CAM this annotation would have to be omitted? Just putting this out there as a prescribed way to remove the 'activated_by' This is the first one I checked where binding has been more addressed and shown.

[pfey03]

Just a note: Bob and I agreed to replace activated_by with has_input for protein binding annotations. These are binding annotations to Dicty calmodulin calA in the table and some others. While we liked activated_by as it more described what calcium 2+ does, we decided to be ok with has_input for these.

[pfey03]

dgcA (DDB0185528) the diguanylate cyclase, has that activity GTP I changed to has_input as GTP is converted to c-di-GMP in the reaction Mg2+ I left with activated_by as it's necessary for reaction to work, and there are putative Mg binding sites, but binding hasn't been shown and I left for now analog to the above.

[pgarmiri]

Hi, CAFA and UniProt annotations have been checked. I have recorded the actions taken in the spreadsheet. In some cases I could delete it completed, in some others I have replaced it with AE happens_during with a 'response to...' term.

The last one I checked it was an enzyme that was activated by calcium (PMID:21584310). I have replaced the AE with happens_during(GO:0071277). Would that be okay or better to request a new terms? (BP 'calcium-dependent citrullination' and MF 'calcium-dependent protein-arginine deiminase activity')

Thanks, Penelope

[BarbaraCzub]

ARUK-UCL updated cc @RLovering

[BarbaraCzub]

ParkinsonsUK-UCL updated cc @RLovering

[pgaudet]

@pfey03 @rjdodson

For

DDB_G0272004 | p2xA |   | GO:0015278 | calcium-release channel activity

How about if I create a child term, "ATP-sensitive calcium-release channel activity" (similar terms exist:
'cADPR-sensitive calcium-release channel activity' 'calcium-induced calcium release activity' 'inositol 1,4,5-trisphosphate-sensitive calcium-release channel activity' 'NAADP-sensitive calcium-release channel activity' 'ryanodine-sensitive calcium-release channel activity'

Does that remove the need for the 'activated by' extension ?

Thanks, Pascale

(I am also looking into the other problematic terms - but I will try to propose the same solution everywhere)

[pfey03]

@pgaudet

Yes "ATP-sensitive calcium-release channel activity" would solve the activated_by for p2X Ca channels. However, I wonder if creating new terms is really so much better than leaving a nice extension.

Also, what to do about "diguanylate cyclase activity" activated by Mg2+. Also adding a new term? I really would dislike using 'has_input' for activating chemicals or other molecules as it looses information.

Thanks!

[pgaudet]

Hi Petra,

After discussion with @thomaspd , @vanaukenk and @ukemi we prefer new terms for the gated channels, as this already exists in the ontology. If you want to capture more explicitly, the GO-CAM "compliant" annotation would be to capture:

• GeneProduct 1 enables binding has_input some calcium directly_positively_regulates ion channel activity (enabled by GeneProduct1)

In P2GO we need to annotate the second enables as a separate annotation.

Does that capture the information correctly?

[pgaudet]

@pfey03 Where is the annotation to 'diguanylate cyclase activity' ? I cannot find it in the table.

@pgarmiri for the citrullination it looks like calcium is more of a co-factor than regulated by calcium, isn't it ? If that's the case then I would no create new terms. We can always add in the definition that the MF ' protein-arginine deiminase activity' is calcium-dependent (see also x-ray structure paper: PMID:27393304). There doesn't seem to be both calcium-dependent and calcium-independent reactions, so there is no need for two terms.

Let me know if that answers your question.

Thanks, Pascale

[pfey03]

the dictyBase lines have been deleted so I cannot lookup anything anymore. Went to back issue and on Jan 15 @pgaudet you seem to have removed them? Not sure how to continue now as I have so many different things to work on. Please restore, thanks!

[pfey03]

ok, I could not add the dicty annotations back but I could save them for myself in Excel. I think I don't have privileges to paste content into the google table. So I'm doing it offline then and comment here unless someone can restore dicty annotations in the current table. thanks!

[pfey03]

@pgaudet - gene dgcA: dictyBase | DDB_G0283453 | DDB0185528 | GO:0052621 | diguanylate cyclase activity | PMID:22864416 | IDA |   | F | Uncharacterized protein | DDB0185528 | protein | taxon:44689 | 20180518 | dictyBase | activated_by(CHEBI:18420 magnesium(2+)

[pfey03]

solve it by adding a child to GO:0052621 diguanylate cyclase activity ? New: magnesium activated diguanylate cyclase activity

[pgaudet]

@pfey03 I dont see where the Mg is needed. They use Mg2+-GTP, which as far as I understand is likely to be the form of GTP that all GTPases use.

Did I miss something ?

[pgarmiri]

@pgarmiri for the citrullination it looks like calcium is more of a co-factor than regulated by calcium, isn't it ? If that's the case then I would no create new terms. We can always add in the definition that the MF ' protein-arginine deiminase activity' is calcium-dependent (see also x-ray structure paper: PMID:27393304). There doesn't seem to be both calcium-dependent and calcium-independent reactions, so there is no need for two terms.

Let me know if that answers your question.

Thanks, Pascale

Hi Pascale, I am happy with your suggestion of including the calcium-dependency in the definition. Let me know when this is done and I will remove the extension(s). Thanks, Penelope

[pgaudet]

@pgarmiri I edited the definition for 'GO:0004668 protein-arginine deiminase activity'

https://github.com/geneontology/go-ontology/issues/18666

[pgaudet]

@pfey03 'GO:0140417 ATP-sensitive calcium-release channel activity' was created here:

https://github.com/geneontology/go-ontology/issues/18649

[pgarmiri]

Thank you Pascale. The extension has been removed. Penelope

[pgaudet]

Great !

[pfey03]

@pgaudet

1. Thanks for adding GO:0140417 ATP-sensitive calcium-release channel activity!

2. diguanylate cyclases. They only occur in bacteria before they were found in all Dictyostelids. I investigated a bit. In bacteria, they need Mg or Mn or even other divalent metal cations, but these are absolute requirements. Not sure all G-proteins require metal cations. https://www.brenda-enzymes.org/enzyme.php?ecno=2.7.7.65 https://enzyme.expasy.org/EC/2.7.7.65 Maybe the definition of GO:0052621: diguanylate cyclase activity should be updated. I might also ask to update the enzyme records above to include something about the occurrence in eukaryotes. Current Def: Catalysis of the reaction: 2 GTP = cyclic di-3',5'-guanylate + 2 diphosphate + 2 H(+). New Def: Catalysis of the reaction: 2 GTP = cyclic di-3',5'-guanylate + 2 diphosphate + 2 H(+); requires divalent metal cations, e.g Mg2+, as cofactors.

If this Def update can be done then I will be done with this review and delete the extension. Thanks!!

[pfey03]

@pgaudet Sorry I fell off this. I finally removed activated_by from ptx channels and annotated to new term (6 annotations). Asking for more GO terms to fix the others . From P2GO there are only 7 more to activated_by

[pfey03]

ok, changed all except the two for the PDE3 where I asked for a new term https://github.com/geneontology/go-ontology/issues/21586

[RLovering]

Hi looking at the annotations that I still need to revise it looks like these were waiting for a decision on how to capture the activated by a chemical (for example).

Please confirm that I should change:

P10276 | RARA enables GO:0004879 | nuclear receptor activity AE activated_by(CHEBI:26536 retinoic acid)

TO P10276 | RARA enables GO:0004879 | nuclear receptor activity AE part_of GO:0048384 retinoic acid receptor signaling pathway

Thanks

Ruth

[pgaudet]

@RLovering

This looks good. In addition if you want to describe retinoic acid you can use 'has part' retinoic acid binding'.

Thanks, Pascale

[pgaudet]

@pfey03 Is this enzyme ~Ca~ divalent cation-activated? In UniProt, ~calcium~ a divalent metal cation is described as a co-factor. https://www.uniprot.org/uniprot/B0G0Y8

Thanks, Pascale

[pfey03]

@pgaudet The PDE3 is NOT Ca- regulated. They tested that, this enzyme is Mn2+ and to a lesser degree Mg2+ activated, PMID:11171061 It's even described in abstract

Also the UNiProt, by similarity says: "Binds 2 divalent metal cations per subunit. Site 1 may preferentially bind zinc ions, while site 2 has a preference for magnesium and/or manganese ions"

[pfey03]

from the abstract PMID:11171061: "The enzyme requires bivalent cations for activity; Mn(2+) is preferred to Mg(2+), whereas Ca(2+) yields no activity."

[Antonialock]

I don't think there is anything for me to do here? It looks like the outstanding ones are BHF-UCL? @RLovering

[vanaukenk]

@RLovering @pgaudet Is it clear what we need to do with the remaining nine annotations? The MF terms don't yet have a logical definition in the ontology, but for now, is it okay to capture in AEs 'part of' the relevant signaling pathway and/or 'has part' the relevant molecule binding term?

[pfey03]

@pgaudet For the PDE3 I now added cation binding has_input Mn2+ and Mg2+ but it's not satisfactory. Something we need to fix when moving all to GO CAM.

[pgaudet]

Hi,

The wikipedia page on cofactors makes a distinction between metal ions that are part of enzymes (magnesium, manganese, cobalt, copper, zinc, and molybdenum), while for calcium, it says

Calcium (...) is needed for the full activity of many enzymes, such as nitric oxide synthase, protein phosphatases, and adenylate kinase, but calcium activates these enzymes in allosteric regulation, often binding to these enzymes in a complex with calmodulin. Calcium is, therefore, a cell signaling molecule, and not usually considered a cofactor of the enzymes it regulates.

So, I have no problem with 'calcium-activated x activity', however for others, it seems like it's just part of the activity, and could be specified in the term definition, but it doesn't justify a distinct GO term.

Also. we have no other activity which is Mn or Mg-dependent. I think in this case we are a bit misled by the phrasing the authors used in the paper

The enzyme requires bivalent cations for activity; Mn(2+) is preferred to Mg(2+), whereas Ca(2+) yields no activity.

That doesn't imply that it's activated by Mn/Mg.

What do you think?

Thanks, Pascale

[pfey03]

I deleted the activated by, as I said, will see what I can do in GOCAM. Also, the human PDE9A binds Mg and Zn. This is the most similar human PDE to the Dicty PDE3

Finally, they showed that the enzyme activity is highest with Mn 2+ and high with Mg 2+ and not existent without cation or Ca2+

[pgaudet]

There are 48 annotations left in QuickGO (may be a little bit behind)

https://www.ebi.ac.uk/QuickGO/annotations?extension=activated_by(chebi

[pfey03]

Not more than 47 as Dicty has none left just didn't make it into prod list yet

[Antonialock]

I removed the CAFA annotation, and the UniProt anntoation extension

[srengel]

3 last remaining SGD activated_by have been removed. just missed them before somehow.

[pfey03]

The dicty pink highlighted ones have all be done before.

@pgaudet about the GO:0052621 diguanylate cyclase activity: There is a link to EC:2.7.7.65 that has Mg2+ as cofactor listed. Maybe the def could reflect that?

[RLovering]

activated by removed from all UCL annotations listed

[pgaudet]

All annotations have been removed.