branched amino acid metabolism

[ValWood]

reviewed here for fission yeast https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9041340/ useful for modelling the pathway quickly. I have captured already most of the 'causal model' in figure 4.

[dexink]

Working on this now @ValWood

Thoughts with an initial look: Seems like there are some dubious annotations that need cleaned up (e.g., conflicts between pombase annotations vs. other databases), may have some unique branches of canonical pathways missing per KEGG (some enzymatic nodes missing from modules, etc.). STRING network (settings: no text-mining, no co-occurrence, no co-expression; high confidence; expanded ~5 times; permalink to be added) reflects some of these lapses in that nodes not usually expected to be linked (largely of other aa metabolic pathways) are in fact linked. Some of this, I suspect, are due to localizations in the mitochondria, trafficking to/from mitochondria, and how this may have influenced experimental data. Published data linked from GO-annotated genes (i.e., branched-chain amino acid metabolism) seems to indicate loose associations with the proteosome, mitochondrial DNA maintenance, etc., all things that somewhat add to the aforementioned suspicions regarding challenges posed by the localization and types of observations made in support of annotations.

Writing up a synthesis in notes for clean contributions here in the issue page. Will follow up soon.

[dexink]

STRING network permalink: https://version-11-5.string-db.org/cgi/network?networkId=bQVWjyPsQ0YA

[ValWood]

I suspect the annotations https://www.yeastgenome.org/locus/S000004347/go linking ILV5 to genome maintenance are incorect/indirecy (these are phenotypes, not processes, it is unlikely the ILV 5 has a direct role in mitochondrial genome maintenance). Most likely defects cause elevated ROS which affects mitochondrial DNA. Over time most annotations to "mitochondrial genome maintenance" have been removed because they describe phenotypes.

was partly dealt with by this ticket, but still some issues remain https://github.com/geneontology/go-ontology/issues/18417

Similarly for the ubiquitin system. There are 4 annotations. One worm and 3 candida. The worm one is definitely an indirect phenotype caused by the downstream affects of increased ROS/stress. I have queried the worm one in Protein2GO.

[ValWood]

The connection between leu1 and atg 1 is only from genetic interactions. The connection is that amino acid levels are sensors for growth and autophagy via the TOR pathway. This is well-known, but is much higher than "amino acid biosynthesis"

i.e When TORC1 signalling is active Gene expression is activated Cells proliferate (growth) Catabolism (autophagy ) inhibited

When TORC1 signalling is inhibited (by amino acid stress) -proliferation (growth slows, cells divide shorter) -autophagy is positively regulated

String is just picking up genetic manipulation used by researchers to modulate TORC signalling (disturb amino acid biosynthesis) to study TOR outputs (autophagy).

We would not connect amino acid biosynthesis and autophagy directly in GO because they are completely separate modules that affect each other by a very long chain of intermediate events.

The effects of TOR signalling are very broad https://www.pombase.org/reference/PMID:34296454 (at least 584 targets), so it does't make sense to make connections between things way upstream and downstream).

this is a nice general summary of the broader process: https://www.sciencedirect.com/science/article/pii/S1097276518309341

but we have these signalling modules largely curated already

[ValWood]

This is TORC signalling http://www.esyn.org/builder.php?projectid=1396&type=Graph

This is one of the problems with STRING for functional inference, it connects genetic interactions and some are very long-range, it only means that you see a different phenotype from the 2 mutants than for any individually. Just because 2 mutants have a stronger/ different phenotype does not make them closely functionally connected.

[ValWood]

STRING takes us back to the hairball we are trying to tease apart. It might be useful if nothing is known, but these are well known, but separable pathways and connections (for example TOR signalling is completely merged into autophagy the downstream process that is controlled)

[dexink]

Just finished a minor batch of "teasing" (ha), see below (beginning to collect together important findings, summaries); what are your thoughts on this annotation: https://www.pombase.org/term/GO:0009316

According to KEGG, LeuC/LeuD reactions are not present in S. pombe as they appear in C5-branched dibasic acid metabolism (https://www.kegg.jp/pathway/spo00660+SPBC1A4.02c)

[dexink]

<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns="http://www.w3.org/TR/REC-html40">

RE: Transport Restrictions Based on Annotated Transporter Genes

 

• No import/export possible for Valine between extracellular space (ECS) and cytosol
• No import/export possible for Valine between mitochondria and cytosol
• No import/export possible for Leucine between cytosol and vacuole
• No import/export possible for Leucine between mitochondria and cytosol
• No import/export possible for Isoleucine between ECS and cytosol
• No import/export possible for Isoleucine between mitochondria and cytosol

 

Simplified:

BCAA | ECS-Cytosol | Cytosol-Mitochondria | Cytosol-Vacuole -- | -- | -- | -- Leucine | X |   |   Isoleucine |   |   | X Valine |   |   | X

<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns="http://www.w3.org/TR/REC-html40">

 

Despite annotations indicating mitochondrial localization of BCAA metabolism, no genes appear annotated for transport of BCAAs into/out of mitochondria…

 

 

 

Gene | Function | Pombase ID | COG -- | -- | -- | -- eca39 | Valine, Isoleucine, Leucine biosynthesis (aminotransferase) | SPBC428.02c | COG0115 hom2 | Isoleucine biosynthesis (aspartate semialdehyde dehydrogenase) | SPCC1827.06c | COG0136 ilv1 | Valine, Isoleucine, Leucine biosynthesis (acetolactate synthase catalytic subunit) | SPBP35G2.07 | COG0028 ilv3 | Valine, Isoleucine, Leucine biosynthesis (dihydroxy-acid dehydratase) | SPAC17G8.06c | COG0129 ilv5 | Valine, Isoleucine, Leucine biosynthesis (acetohydroxyacid reductoisomerase) | SPBC56F2.12 | COG0059 ilv6 | Valine, Isoleucine, Leucine biosynthesis (acetolactate synthase regulatory subunit) | SPBC14C8.04 | COG0440 mmf1 | Isoleucine biosynthesis (mitochondrial matrix protein, YjgF family protein Mmf1, reactive intermediate imine deaminase A homolog) | SPBC2G2.04c | COG0251 mmf2 | Isoleucine biosynthesis (mitochondrial matrix protein, YjgF family protein Mmf2, reactive intermediate imine deaminase A homolog) | SPAC1039.10 | COG0251 tda1 | Isoleucine biosynthesis (threonine ammonia-lyase) | SPBC1677.03c | COG1171 leu1 | Leucine biosynthesis (3-isopropylmalate dehydrogenase) | SPBC1A4.02c | COG0473 leu2 | Leucine biosynthesis (3-isopropylmalate dehydratase) | SPAC9E9.03 | COG0065 leu3 | Leucine biosynthesis (2-isopropylmalate synthase) | SPBC3E7.16c | COG0119 dao1 | D-valine metabolism (D-amino acid oxidase) | SPCC1450.07c | COG0665 fnx1 | Isoleucine vacuolar transport (vacuolar amino acid transmembrane transporter) | SPBC12C2.13c | KOG0254 fnx2 | Isoleucine vacuolar transport (vacuolar amino acid transmembrane transporter) | SPBC3E7.06c | KOG0254 agp3 | Leucine plasma membrane transport (plasma membrane leucine transmembrane transporter) | SPCC965.11c | COG0531 vba2 | Valine vacuolar transport (vacuolar amino acid uptake transporter) | SPBC460.03 | KOG0254 pub1 | Regulation (negative), Leucine import across plasma membrane (HECT-type ubiquitin-protein ligase E3) | SPAC11G7.02 | COG5021 pas1 | Regulation (positive), Leucine import across plasma membrane (cyclin) | SPAC19E9.03 | KOG1674 tsc1 | Regulation (positive), Leucine import across plasma membrane (Rheb GAP, hamartin) | SPAC22F3.13 | ENOG502CP4E tsc2 | Regulation (positive), Leucine import across plasma membrane (Rheb GAP, tuberin) | SPAC630.13c | KOG3687 snr1 | Valine catabolism (mitochondrial ribosomal protein subunit S47/3-hydroxyisobutyryl-CoA hydrolase) | SPBC2D10.09 | COG1024

Genes extracted from Pombase by GO term parentage re. valine/isoleucine/leucine metabolism and transport

[dexink]

More notes (how I arrived at the above deficits):

// using STRING to play with data, stringently filtered, alongside what is published in KEGG, cross-referenced with what is currently in Pombase:

<html xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:dt="uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" xmlns="http://www.w3.org/TR/REC-html40">

With regulators removed (i.e., pub1, pas1, tsc1, tsc2), STRING network generated with following settings:

Confidence | Edges | Text-mining | Experiments | Databases | Co-expression | Neighborhood | Gene Fusion | Co-occurrence | Shell 1 | Shell 2 -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- High (0.700) | Evidence | no | yes | yes | no | yes | yes | no | <=20 | <=10

*S. pombe systematic identifiers used for network initiation (STRING option "Multiple Proteins")

 

https://version-11-5.string-db.org/cgi/network?networkId=bzHlnZg7axuU

Applied kmeans clustering (6 clusters):

Changing confidence to "medium" (0.400) and applying kmeans clustering (8 groups, no edges between clusters): https://version-11-5.string-db.org/cgi/network?networkId=bG3j4qcMNyMp

Manually re-organizing the network's clusters by likeness:

Separation of the ilv1/ilv3/ilv5/ilv6 genes is noted; these enzymes are annotated as being involved in the synthesis of each of the three BCAAs. Genes ilv5 and ilv3 (SPAC17G8.06c.1) grouped together, and separately ilv1 and ilv6 (SPBC14C8.04.1).

Re-examining KEGG's documentation of BCAA biosynthesis:

We can see that D-erythro-3-Methylmalate appears to be a dead- or loose-end in this pathway for S. pombe…

Expanding the view of the preceding pathway, KEGG only lists Ilv6 and Leu3 within this pathway of S. pombe:

[dexink]

I should've said more clearly before, too, that I am just using STRING to visually organize and re-group published data as it relates to things we already have curated, etc. Just keeping it all organized

[ValWood]

I agree, https://www.pombase.org/term/GO:0009316 Is incorrect mapping, probably OK for bacteria, but leu2 and leu3 are in different compartments in yeast it seems:

Leu2 3-isopropylmalate dehydratase activity https://www.yeastgenome.org/locus/YGL009C#go cytosolic

leu3 2-isopropylmalate synthase Mitochondrial

This should be reported here with an "InterPro" tag https://github.com/geneontology/go-annotation/issues

I will filter it from PomBase.

[ValWood]

Would be interesting to link up the amino acid transporters with substrates/ directions/ locations. I never felt very confident to do this and nothing much is really published for yeast....also difficult to assess specificities because of lots of duplication and divergence.

[dexink]

If it exists, I will try to find it. I suspect other dimorphic fungi may have some thing published for their yeast forms relating to this, but it is on my list of possible pools of info to tap into. I very much doubt that the well would be dry for relevant experimental data considering the amount of money in eukaryotes (i.e., humans) dependent on BCAAs.

[dexink]

I agree, https://www.pombase.org/term/GO:0009316 Is incorrect mapping, probably OK for bacteria, but leu2 and leu3 are in different compartments in yeast it seems:

Leu2 3-isopropylmalate dehydratase activity https://www.yeastgenome.org/locus/YGL009C#go cytosolic

leu3 2-isopropylmalate synthase Mitochondrial

This should be reported here with an "InterPro" tag https://github.com/geneontology/go-annotation/issues

I will filter it from PomBase.

Also, I am less familiar with the expected reporting formats for the GO issues lists (e.g., priorities, etc.). I will check into the context of the issue before posting.