Insulin Processing (R-HSA-264876)

[ukemi]

• [x] Make mouse model gomodel:62f58d8800001500
• [x] The chain of events in the import seems broken after the last protease step. We need to investigate this. Looks like a bad blurb - the next-to-last cleavage generates mature insulin + a C peptide precursor. The last annotated cleavage removes two c-terminal residues from the precursor to yield C peptide - now fixed in blurb and annotation (add missing input waters and output free amino acids).
• [x] Is there an intervening reaction missing in the human pathway catalyzed by P4HB? New evidence from mouse that suggests an update to the model and the blurb YES - current evidence suggests that P4HB mediates proinsulin folding and disulfide bonding and itself gets reduced in the course of the reaction. It is then re-oxidized (its own disulfide bond is reformed) in a reaction catalyzed by ERO1B - see revised reaction R-HSA-264997 and new reaction R-HSA-9817575. Now fixed? As a bonus, the new references include evidence for mouse proteins.
• [x] The transporters that mediate zinc uptake into the Golgi are wrong according to more recent work. Now fixed? R-HSA-437139 and R-HSA-437129
• [ ] And PMID: 30230185 for evidence that there is a post-translational path for import of proinsulin into the ER as well as a co-translational one - defer this to pathway revision project?

[ukemi]

P4HB
I've also got a step catalyzed by Hsp90b1 as the chaperone: PMID: 30670477 I've got Ero1b upstream of the P4hb, but since it's a cycle I think this is consistent. I actually think you are more accurate. I've still got different transporters for the Zinc than you do. PMID: 12446736, PMID: 11997387

[deustp01]

it's a cycle

Tangential but important and complicated rabbit hole - many model-building strategies, I am told, just can't handle cycles because the cycle gets turned into an infinite loop with no escape. Do issues like that arise with GO-CAM?

[ukemi]

Not according to what I have been told.

[ukemi]

I went back and changed the transporters in the mouse model to 5 and 8. This is similar to what you have, but I don't see evidence for 8 being in the Golgi in the mouse, so there is still that difference in our models. I removed the other two transporters that I had in the mouse model due to lack of evidence of their being in the beta cells.

[deustp01]

I will re-check the evidence for human 8 in human Golgi.

Here's the re-check result. The Kambe et al. 2021 review PMID: 33485965, Figure 1a

shows ZnT8 associated with Golgi vesicles and also with insulin granules, but (as far as I can see - time for new eyeglasses?) shows ZnT5 only as a heterodimer with ZnT6 and only associated with the Golgi, not with insulin granules.

Kambe et al. 2001 describe the localization of ZnT5 by immunofluorescence and immuno-EM as being only with secretory granules (Figure 4) but then in Figure 5 show ZnT5 localizing to the Golgi, albeit in cells in which the ZnT5 construct is overexpressed, in contrast to what I see in the later review cartoon.

Bottom line: evidence, more or less for both ZnT5 and ZnT8 in Golgi in a more or less human system; from the cartoon in Kambe et al. 2021 only ZnT8 is present in secretory granules, so when I get that far in the pathway I should delete the Reactome ZnT5 / secretory granule reaction, and leave only the ZnT8 one