Repository navigation
Fixing collagen synthesis #335
Description
Description of the issue:
All reactions pertaining to collagen synthesis in v1.10 (MAR06983, MAR06984, MAR06985, MAR09553) are dead-end reactions. Adding an export reaction to remove 1,2-D-glucosyl-5-D-(galactosyloxy)-L-lysine-procollagen[c] enables the reactions to carry flux, except for MAR09553 which seems to be a duplicate of MAR06983. Furthermore, collagen synthesis also involves proline residues, which are not captured in the rxns in the current HUMAN-GEM.
Expected feature/value/output:
Collagen production and secretion is an important feature of cells such as fibroblast, and it can be an important sink of aminoacids in such cells. Since other extracellular matrix components like hyaluronate can be produced by the model, I believe that curating and improving this set of reactions can be of value to the community. I am creating this issue in case someone else has thought of this problem or if anyone has any ideas on how to address it. I will start by looking at how other GEMs deal with collagen synthesis.
I hereby confirm that I have:
- Checked that a similar issue does not exist already
Activity
haowang-bioinfo commentedon Nov 16, 2021
@CadavidJoseL very nice input.
can you please look into if there is any transporter(s) that assist the secretion of
1,2-D-glucosyl-5-D-(galactosyloxy)-L-lysine-procollagen?CadavidJoseL commentedon Nov 16, 2021
@Hao-Chalmers I'll have to look at this pathway in detail, it is more complicated than I thought. Procollagen seems to be exported in vesicles and it matures fully outside the cell. Glycosylation seems to be one of the final steps in the synthesis process, so for practical purposes maybe we consider
1,2-D-glucosyl-5-D-(galactosyloxy)-L-lysine-procollagenas the final collagen precursor to be exported. I am looking at fixing other gene associations in the HUMAN-GEM, and also at whether we can improve the collagen synthesis pathway in general, but I'll be a bit slow given my current time constraints.CadavidJoseL commentedon Nov 16, 2021
More generally, I found this paper (with input from your group) that looks at incorporating protein export. Maybe this is an even better thing to integrate?
mihai-sysbio commentedon Feb 17, 2022
@CadavidJoseL did any idea crystallize?
CadavidJoseL commentedon Feb 17, 2022
@mihai-sysbio other than my previous comment, I haven't been able to spend too long on this since I am trying to get experiments to graduate this year. Terribly sorry about this. If we want to address collagen synthesis only, that might be simpler, but if we want to incorporate more secreted proteins, it's much longer. If you have an idea of how to approach the problem systematically, I am happy to help with parts of it!
edkerk commentedon Jul 11, 2026
Picking this up. The two concrete problems are now fixed so the collagen synthesis pathway can carry flux:
MAM02768) was itself a dead-end input, so it could never carry flux. MAR06983 uses the properly-sourced[protein]-L-lysine. The orphanedMAM02768is removed too.MAM00232(1,2-D-glucosyl-5-D-(galactosyloxy)-L-lysine-procollagen). Added the extracellular metabolite, a transport to[e], and an "Exchange of …" reaction, modelled like the existing hyaluronate secretion, so this final precursor can be secreted and the pathway carries flux.This treats the glucosylgalactosyl-hydroxylysine-procollagen as the final secreted precursor, as you suggested (procollagen is exported in vesicles and matures extracellularly).
On the broader ideas, a general secreted-protein-export framework and collagen proline hydroxylation are intentionally left out. Human-GEM is a generic human model, a template of the metabolic capabilities shared across cell types rather than a cell-type-specific reconstruction. A full protein-secretion framework (the Nature 2019 approach you linked) and the residue-level detail of collagen maturation are specialised, cell-context-dependent additions that belong in context-specific or dedicated protein-secretion models built on top of Human-GEM. The generic model only needs the existing collagen pathway to be functional and to act as the amino-acid sink you noted, which this change provides.
feat: enable collagen secretion and remove duplicate MAR09553 (#1043)
edkerk commentedon Sep 28, 2026
Collagen lysine hydroxylation and glycosylation (MAR06983-MAR06985) are modelled in the cytosol, but lysyl hydroxylase and COLGALT1 act in the ER lumen (PMID:7982970, PMID:20470363). Moving them to the ER needs [protein]-L-lysine[r], three new ER metabolites, and a route for 2-oxoglutarate and succinate in the ER (the ER prolyl 4-hydroxylase MAR01649 is blocked for the same reason). The secretion reaction MAR20195 (#1043) now has a note saying it stands in for secretory-pathway export.
edkerk commentedon Sep 28, 2026
Follow-up to #1043, in #1105 (open): MAR20195 (procollagen secretion, cytosol to extracellular) now has references and a note that it stands in for secretory-pathway export; MAR06983 and MAM00204c record the merged duplicates MAR09553 and MAM02768c. Details and evidence are in the comment on #1043.
edkerk commentedon Oct 5, 2026
Fixed in Human 2.1.0.