Heparan sulfate biosynthesis (R-HSA-2022928)

[ukemi]

• Create a mouse model based on the Reactome import.
• Should R-ALL-744229 N-acetyl glucosamine map to CHEBI:57705?
  NO - R-ALL-744229 is the set of UDP-N-acetyl-alpha-D-glucosamine(2-) (ChEBI:57705) OR UDP-D-glucose(2−) (CHEBI:58367) (was UDP-D-glucose (ChEBI:18066) - wrong charge at pH 7.3, now fixed). Does it make biological sense to group these two chemicals? The issue is FIXED here by splitting the reaction in two, one for the transport of each UDP-sugar and only UDP-glucosamine transport is linked to the HS biosynthesis pathway.
• That raises a general issue - DEFER for future. Reactome has no chemical-similarity standards for grouping entities into a Reactome set, only the requirement that they be functionally interchangeable in the events that involve the set. Further, set membership can be changed at any time so even if all current members are similar enough to be grouped and a grouping term exists in ChEBI, there is no guarantee that this grouping will persist, so trying to assign a grouping ChEBI ID to a Reactome set is complicated and risky. Future cleanup: it would be great to have a script that could take a reaction in which a set of inputs is converted to a set of outputs and compose all the individual reactions implied by the set (input 1 to output 1, input 2 to output 2, etc.) on the fly for purposes, e.g., of alignment to ChEBI, all of this coupled to more conservative rules in Reactome for set membership.
• R-HSA-2022928 maps to [GO:0006024] glycosaminoglycan biosynthetic process, why not [GO:0015012] heparan sulfate proteoglycan biosynthetic process? The more specific process term looks OK, so changed / fixed in Reactome. BUT as far as I can tell, heparan sulfate proteoglycan is a kind of glycosaminoglycan, so do we need a GO:0015012 is_a GO:0006024 parent - child relationship in the ontology, and if not is the change correct? Check for chemical sanity and open a GO ontology ticket to add relationships? @ukemi
• Merge R-HSA-2076383 and R-HSA-2076611? No. There is a family of sulfottransferases, HS3ST1, 2, 3A1, 3B1, 4, 5, and 6. Reactome asserts that HS3ST1 is localized to the Golgi lumen while all the others are localized to the Golgi membrane. The evidence for this is that in UniProt all but HS3ST1 are said to have membrane-spanning domains while HS3ST1 lacks such a domain (a stealth NOT annotation, but never mind). Reactome has a set of 1 and possibly 5 catalyzing one of these reactions, while a set of 2, 3A1, 3B1, 5, 6 and possibly 4 catalyze the other. Clean-up (done): replace the set of 1 and possibly 5 with just the EWAS 1, to be the Golgi-lumen catalyst, while members of the other set provide the Golgi-membrane catalyst. And edit the pathway diagram to put the HS3ST1 icon in the Golgi lumen (it’s now in the membrane).
• Some HS molecules are secreted outside the cell and some go to the plasma membrane. Do we need separate pathways for each? Maybe not if we decide to split out the secretion into a separate pathway? PD opinion - NO.
  Here's the summation / blurb for the final step of the pathway, Some HSPGs are secreted to the plasma mambrane" - "Depending on the nature of the core protein HS-GAGs are attached to, they will either translocate to the cell surface or be secreted into the extracellular matrix (ECM). Here, HS-GAGs are shown to translocate to the cell surface (Kjellen & Lindahl, 1991). The mechanism of transfer from the Golgi apparatus to the cell surface and beyond is unknown but most likely involves the trans-Golgi network." This suggests two reasons NOT to split secretion from membrane targeting. The first mechanical one is that we don't really know how the cell makes this distinction so we don't have the basis for identifying distinct separate biological processes. Second and more basic, do we want / need this distinction, or is it OK to say a synthesis / maturation process ends when the end product is delivered to its destination. Here both destinations must involve some amount of transport so the distinction between the two locations / fates is pretty limited.
• Relationship between heparan sulfate proteoglycan (GO:0015012) and heparin biosynthesis (GO:0030210)? Can we disentangle this, is it just the extent of iterative sulfation? From PMID: 22566219 and PMID: 27900574, cleanly disentangling looks hard - mostly quantitative differences so we can capture all relevant gene products and their molecular functions without disentangling.
• Shouldn’t R-HSA-2024100 and R-HSA-2076371 be mapped to heparosan-N-sulfate-glucuronate 5-epimerase activity, GO:0047464? Yes, done. Comment: we already had the identical gene product + GO MF associated with both reactions. It looks like the old MF term might apply to epimerization of UDP-glucuronate, but not to glucuronate residues in an oligomer.
• We should try to map the subpathways. See the children of GO:0015012. Or would they not be considered their own processes any more? In Reactome, we could leave the pathway labeled with GO:0015012 and label individual reactions in the pathway with GO:0015015 or GO:0015014 according to whether they involve modification of existing saccharide residues or addition of new ones, respectively. Would that work? Tangential question: R-HSA-2022854 keratan sulfate biosynthesis keratan sulfate biosynthesis (R-HSA-2022854) #168 also involves both addition and modification steps, but GO:0018146 keratan sulfate biosynthetic process has no children to make this distinction. Should we request new GO terms? @ukemi
• Charge state on the UDP-D-glucose in R-HSA-744231? Wouldn’t the phosphates be deprotonated? CHEBI:58367 - yes, done (3.a.ii above)
• Are we missing a bunch of preceding reactions that can get the various UDP-’sugars’ into the Golgi? Some of them are preceding events and some are not even though they catalyze the transport of the correct molecules. R-HSA-727802.
• First, a charge-state side rabbit hole:
  • - UDP-N-acetyl-D-galactosamine (CHEBI:16650) should be UDP-N-acetyl-D-galactosamine(2−) (CHEBI:57847) done
  • - UDP-α-D-glucuronic acid (CHEBI:17200) should be UDP-α-D-glucuronate(3−) (CHEBI:58052) done
  • - UDP-α-D-xylose (CHEBI:16082) should be UDP-α-D-xylose(2−) (CHEBI:57632) done
  • - aldehydo-L-iduronic acid (CHEBI:28481) should be aldehydo-L-iduronate (CHEBI:21338) done
  • - CMP-N-acetyl-β-neuraminic acid (CHEBI:16556) should be CMP-N-acetyl-β-neuraminate(2−) (CHEBI:57812)
• Now, a list of reactions that move relevant molecules into the Golgi lumen: R-HSA-735702 UDP-Gal, UDP-GalNAc; R-HSA-741450 UDP-GlcNAc; R-HSA-742354 UDP-GlcNAc; R-HSA-742373 UDP-Xyl; R-HSA-744231 UDP-Glc, UDP-GlcNAc; R-HSA-742345 GDP-Fuc; R-HSA-741449 PAPS; R-HSA-727807 CMP-Neu5Ac
• Leads to a catalog of glyco-synthetic reactions that consume each: 727815 CMP-Neu5Ac input for 2046285; 735682 UDP-Gal input for 1889978, 1889981, 2046298, 2025723, 2046265; 735700 UDP-GalNAc input for 1971487, 9632033, 1971482; 914003 UDP-GlcNAc input for 2025724, 2022851, 2022919; 742361 UDP-Xyl input for 1878002 (already done); 741440 PAPS input for [long list - fixed]