Signalling by insulin receptor (R-HSA-74752 )

[ukemi]

• [x] Make mouse models
• [x] Pathway boundary question- Would R-HSA-74718 fit better under R-HSA-77387? The internalization isn't really part of the signaling is it? It doesn't connect to anything downstream other than the recycling steps, so is it a signal?
• [x] Make R-HSA-74716 a receptor ligand activity (GO:0048018)? Is it possible to make insulin the enabler? For the receptor activity (R-HSA-74715), is it possible to make the INSR the enabler? Interesting discussion for the weeds.
• [x] Change mapping of (R-HSA-74715) from 'transmembrane receptor protein tyrosine kinase activity' to insulin receptor activity (GO:0005009).

Detailed comments below - conclusions are: R-HSA-74718 / boundary issue - resolved, I think R-HSA-74716 - current Reactome data model doesn't allow this R-HSA-74715 - MF term changed to GO:0005009

[deustp01]

• Pathway boundary question-

Maybe not. Here's what the summation blurb for R-HSA-74718 says

Almost concomitantly [with receptor autophosphorylation, the preceding step in the pathway] a second effect resulting from the tyrosine phosphorylation of the insulin receptor begins to occur. The phosphorylation of the tyrosine in the NPEY sequence found in the juxtamembrane is also a signal for endocytosis to occur. Whilst invagination of the plasma membrane commences the receptor tyrosine kinase activity continues unabated as does substrate phosphorylation.

As the invagination continues certain proteins are concentrated in the area of invagination. In addition to the insulin receptor itself there is a recruitment of insulin-specific protein tyrosine phosphatases (PTPs). This process takes less than one minute. (The identity of these PTPs is not clearly established yet.)

The formation of the endosome containing the activated ligand-receptor complex is completed within two minutes following ligand presentation at the plasma membrane and is maximal by five minutes. Endocytosis of activated receptors has the dual effect of concentrating receptors within endosomes and allowing the insulin receptor tyrosine kinase to phosphorylate substrates that are spatially distinct from those accessible at the plasma membrane. The endosome also contains other proteins crucial to the signal transduction process. These include a proton pump and the insulin degrading activity. It is not certain how these proteins arrive in the endosome since it could be via the endosome maturation or fusion pathways.

This makes the internalization step look like it's augmenting / modulating / propagating the signaling process. If that is right, then this event is in the right place, though I would change the order to put it before rather than after the "Insulin receptor signalling cascade" pathway in the event hierarchy -

[ukemi]

So based on what you say above, that makes sense. So in GO-CAM speak, the internalization would be embedded between the autophosphorylation and the cascade? Am I understanding that? It seems reasonable to me.

[deustp01]

the internalization would be embedded between the autophosphorylation and the cascade?

Yes - I have re-done the event order in Reactome, and added R-HSA-74718 "Internalisation of the insulin receptor" as a precedingEvent to R-HSA-74726 "Disssociation of insulin from insulin receptor".

[ukemi]

It should fall out into the GO-CAM. Is there any way you can add 'receptor internalization' to the black box reaction while you are at it?

Done - R-HSA-74718 "Internlisation of the insulin receptor" now has GO:0031623 "receptor internalization" as its GO biological process attribute

See what I did here so far based on this discussion: http://noctua.geneontology.org/editor/graph/gomodel:62b4ffe300004795 Next I'll make the downstream events for the different flavors.

[deustp01]

• Make R-HSA-74716 a receptor ligand activity (GO:0048018)? Is it possible to make insulin the enabler?

Now, neither is possible. As an editorial policy we prohibit creation of catalystActivity instances whose molecular function attribute is anything except an instance of catalyst activity or transporter activity, so we can't associate receptor ligand activity (GO:0048018).

In the future, though, nothing in the data model prohibits creating catalystActivity instances with other kinds of molecular function attributes. In fact, we have created ones with binding activity in the past and nothing broke in the data base or the web display. If we go down this road, I would want to create a new class, bindingActivity, with attributes GO molecular function (required) and participating entities (the inputs that assemble into the output complex).

• For the receptor activity (R-HSA-74715), is it possible to make the INSR the enabler?

Now, no. In the future, we would need an "enabler" attribute, perhaps attached to the event class, to hold this information, as well as rules for how to use it in curation. Thinking more about "enabling" in binding reactions, I had previously assumed that one of the inputs could reliably be identified as the enabler of the interactions to form a complex, but thinking more, starting with R-HSA-74715, we probaby need to allow for entities other than inputs to serve as enablers, so "enabler" becomes an attribute of the reaction (like regulation).

[ukemi]

Now, neither is possible. As an editorial policy we prohibit creation of catalystActivity instances whose molecular function attribute is anything except an instance of catalyst activity or transporter activity, so we can't associate receptor ligand activity (GO:0048018).

An alternative would be to write a rule for the converter. If a molecular event has two inputs and directly_positively_regulates a receptor activity that has a catalyst, then the molecular event is a receptor ligand activity enabled by the input that is not the catalyst of the receptor activity.

[deustp01]

• For the receptor activity (R-HSA-74715), is it possible to make the INSR the enabler?

Now, not now. In the future, we would need an "enabler" attribute, perhaps attached to the event class, to hold this information, as well as rules for how to use it in curation

Is there any way you can add 'receptor internalization' to the black box reaction while you are at it?

Yes (done). I can add a biological process attribute to any event. (Remember that when we mine our data for gene product : BP mappings for our GAF, we traverse the event hierarchy bottom up, so if a reaction has its own BP term we do not look at its parent pathway to find a BP term.)

[ukemi]

Yes, I can add a biological process attribute to any event. (Remember that when we mine our data for gene product : BP mappings for our GAF, we traverse the event hierarchy bottom up, so if a reaction has its own BP term we do not look at its parent pathway to find a BP term.)

I think that works out. I'm not sure in our model we would want all the endocytosis genes that might be responsible for the internalization to be grouped with the signaling pathway. This is the advantage of the black box.

[ukemi]

    For the receptor activity (R-HSA-74715), is it possible to make the INSR the enabler?

Now, not now. In the future, we would need an "enabler" attribute, perhaps attached to the event class, to hold this information, as well as rules for how to use it in curation

Sorry, not an enabler, a catalyst activity. That's what we mine to get the GO-CAM enabler. Analagous to what's here R-HSA-163750.

[deustp01]

Sorry, not an enabler, a catalyst activity. That's what we mine to get the GO-CAM enabler.

Right - the hypothetical new bindingActivity class should behave like catalystActivity does now (and we in fact know this because a few catalystActivity instances created in violation of editorial policy were picked up in Ben's early GO-CAM runs and handled correctly)

[deustp01]

• Change mapping of (R-HSA-74715) from 'transmembrane receptor protein tyrosine kinase activity' to insulin receptor activity (GO:0005009).

Problem: we are asserting that the input complex catalyzes its own phosphorylation so in the Reactome view tyrosine kinase activity is correct. More weeds to discuss! BUT in fact insulin receptor activity (GO:0005009) is_a kinase activity so in fact there is no problem and I've made the change. Now a new question. The catalystActivity attribute for that reactions has insulin:insulin receptor as its physical entity (which is right because this complex is both the input for the phosphorylation reaction and also the catalyst of it. Would it move us in a useful direction to identify what you called the beta chain of the cleaved insulin receptor protein and what we call INSR(763-1382) as the active unit of that catalystActivity? I've done it because that in fact is correct Reactome curation practice, but if this confuses things I can un-do it. @ukemi

[ukemi]

Sorry @deustp01. I saw all the tasks were done, so I closed this one a couple weeks ago.

[deustp01]

Re-opened to make one more fix that I overlooked earlier. Found two candidate enzymes to catalyze the dephosphorylation of the insulin receptor in the endosome (R-HSA-74733), so added that information. Also added 12 input H2O's to balance the 12 output phosphates, so the reaction can now "map up" to the generic Rhea reaction (Rhea:10684) for dephosphorylation of a protein tyrosine residue. Note the two possible long-term issues:

• the idea of "mapping up from a specific Reactome reaction to a more general Rhea one (here I expect Rhea will not want to create new reactions because the generic one already fully specifies as much of the chemistry as they want to capture)
• annotation pf proteins with weak or conflicting evidence (here, I think it is unclear how many phosphatases perform this reaction in vivo and whether different ones do it in different tissues or different physiological states). Candidate sets make this easy - do we want guidelines and limitations?