obsolete chromatin silencing / reannotations to deal with

[ValWood]

~chromatin silencing has a parentage GO:0097549 chromatin organization involved in negative regulation of transcription

but this isn't really always true. For constitutive heterochromatin it is a little indirect.

I think it should move to "negative regulation of gene expression" (the same parent as "gene silencing".

However, it would be good to get transcription working group input on this.~

Most of this ticket is out of date. Chromatin silencing will be obsoleted.

[pgaudet]

I looked at this before, and these two terms seem pretty synonymous to me, so I didn't propose a change in ontology structure (but we should consider a merge)

https://en.wikipedia.org/wiki/Gene_silencing

[ValWood]

chromatin and gene silencing are different. At least "gene silencing" covers all mechanisms post-transcriptional, co-transcriptional and chromatin silencing (which is neither). However, we now know more than we did 20 years ago that most of the annotations should move to some sort of "heterochromatin formation" term. These are really phenotypes established in GO before we could be sure of the mechanism. A big clean up required but I think we need experts. .....

Colin might be able to provide input here? At least for the high level which should be related directly to transcription, and which not....

[bmeldal]

Colin is @colinlog (Welcome to the world of GitHub tickets!)

[colinlog]

To me, both facultative and constitutive heterochromatin are forms of chromatin organization of the kind that negatively regulate gene expression. I therefore agree with the proposal that these 'chromatin silencing' annotations should be linked to a term in the ‘negative regulation of gene expression’ branch of GO (I’ll abbreviate this to ‘NRGE’).

In fact, to me ’Chromatin silencing’ is an ambiguous word combination because it is not immediately /intuitively obvious from this terminology that it is gene expression that is silenced rather than another chromatin feature; you could eg. 'silence' the heterochromatin to obtain euchromatin?

The closest term I can think of is indeed heterochromatin formation. NRGE can have three broad sub-classes from a transcriptional perspective; chromatin (pre-transcriptional, DNA-protein-trans-acting-RNA complexes are the actors), co-transcriptional (RNA polymerase Pre-Initiation Complex formation at promoter and onwards is the target) and post-transcriptional (I would include ‘co-transcriptional alternative splicing regulation’ in this category if it results in absence of a protein, as well as regulation of upstream ORF ATG usage leading to absence of protein product or a dominant negative product, a process that is known to play out at the ribosome level- see PMC5016099). In conclusion, I support moving/merging annotations from ‘chromatin silencing’ to ‘heterochromatin formation’. But, only if that is what the results that are annotated there actually describe…

[colinlog]

Sorry did not mean to close...

[ValWood]

@colinlog this is really useful. Hopefully we can follow up on this once the main part of transcription is completed.

[pgaudet]

to be discussed on a transcription call

[mah11]

On a related note, should there be a link between 'heterochromatin assembly involved in chromatin silencing' (GO:0070869) and 'establishment of chromatin silencing' (GO:0006343)?

(note: I added GO:0070869, but it's been a long time since I did, so I don't really remember if the two are unconnected on purpose or just because I didn't know the silencing area well then and still don't now)

happy to move this question to its own ticket if need be

[colinlog]

Hi Midori, Val, Birdit & Pascale,

GO:0070869 Any process that results in the assembly of chromatin into heterochromatin and contributes to chromatin silencing. Source: GOC:mah GO:0006343 The initial formation of a transcriptionally silent chromatin structure such as heterochromatin. Source: GOC:mah

Many of these proteins are A. nidulans annotations. The two lists are relatively short (120 and 21 proteins) and they do not overlap. I therefore guess that the studies these annotations come from are independent, or that one study insisted on making a distinction, that is not immediately obvious to me. Ultimately, it might be a distinction between SIR complexes, H3K9me3 and H3k27me3 modification?

Overall, just on the basis of names and definitions I support merging the two terms, and then to envisage to make 3 or 4 children for the different flavors of heterochromatin to be annotated (H3K9me3 = constitutive heterochromatin to silence transposons, H3K27me3 is facultative heterochromatin to silence developmental regulators, lamin association and the 'rubish bin' 'dark/unmodified/as yet uncharted heterochromatin' that can only annotated to the parent term? Some histone lysin deacetylases would fit all bins?

The heterochromatin systems are more complex than one naively would assume because quite some loci have both K9 and K27-based heterochromatin within one cell population!

For animal gene products, the enzymes that do H3K9me3 and H3K27me3, + the subunits of the complexes they reside in would be annotated to heterochromatin. Nuclear Lamin association factors for the third child? Thomas Jenuwein and many researchers have worked on the mammalian polycomb gene products. For Drosophila, the list of researchers involved over the years is rather long too.

All this will become clearer once we have lists of epigenetic enzymes and complexes that have an ascribed epigenetic Molecular Function -> @bmeldal

Genome-wide studies that could be used to support the DNA centric annotations for Heterochromatin start with Bas van Steensel (Drosophila and human/mouse I believe). Current lists of target loci can be obtained from groups (EBI: Adam Frankish & Daniel Zerbino, others elsewhere) who have performed genome segmentation on the basis of epigenetic modification combinations reported by the ENCODE as well as by the Blueprint consortia.

Hope this helps, Colin

On Mon, Jan 6, 2020 at 12:32 PM Midori Harris notifications@github.com wrote:

On a related note, should there be a link between 'heterochromatin assembly involved in chromatin silencing' (GO:0070869) and 'establishment of chromatin silencing' (GO:0006343)?

(note: I added GO:0070869, but it's been a long time since I did, so I don't really remember if the two are unconnected on purpose or just because I didn't know the silencing area well then and still don't now)

happy to move this question to its own ticket if need be

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/17160?email_source=notifications&email_token=ALZVLKGPZ3EUDQQIMB5J7E3Q4MJFDA5CNFSM4HGIPE2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIFGKEI#issuecomment-571106577, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALZVLKDPJE5MB63BHPQBHFDQ4MJFDANCNFSM4HGIPE2A .

[mah11]

@colinlog - Thanks for all the information! I was already at (perhaps even a bit past) the limit of my understanding when I posted my earlier comment, so I appreciate the input but will have to leave sorting out the ontology to others.

Based on GO and phenotype annotations in PomBase, I would like to double-check whether this distinction:

H3K9me3 = constitutive heterochromatin to silence transposons, H3K27me3 is facultative heterochromatin to silence developmental regulators

... holds for all species, because for fission yeast we don't have anything about H3K27 methylation. We do have some papers* looking at H3K9 methylation in the context of facultative heterochromatin (they often monitor H3K9me2 rather than H3K9me3, but I don't know if this is biologically important or dictated by experimental convenience). Constitutive heterochromatin is found in pericentromeric regions, subtelomeres, and the silent mating-type cassette.

@ValWood, do you know whether H3K27 methylation has been observed in pombe?

*e.g. PMID:31269446, PMID:31260531

[ValWood]

hi, yes https://www.pombase.org/reference/PMID:30773398

[ValWood]

Oh, actually this is H3K37, not 27.I have not come across 27...

[mah11]

yeah, we have nothing annotated to GO:0070734 or any descendant (including the MF GO:0046976), and no phenotypes mentioning H3K27

[ValWood]

At the recent fission yeast meeting a chromatin expert (Rob Martienssen) also thought that "silencing" terms were a bit "outdated" , and that we could move to more "mechanistically" described terms

We describe by phenotype "silencing" mechanism "heterochromatin assembly" region pathway e.g. methylation dependent OR by small RNA (problematic because often both are involved)

and other ways...

[colinlog]

Indeed no H3K27me in Pombe as far as I could see.

The Nature paper PMID: 28682306 https://www.ncbi.nlm.nih.gov/pubmed/28682306 is good for H3K9me2 and me3, which are proposed there to be facultative and constitutive respectively. Awareness of the involvement of the RNAi machinery could (?) have been the initial observation for a distinction between GO:0070869 Any process that results in the assembly of chromatin into heterochromatin and contributes to chromatin silencing. Source: GOC:mah and GO:0006343 The initial formation of a transcriptionally silent chromatin structure such as heterochromatin. Source: GOC:mah

However, in the above Nature paper they argue for RNAi being involved more in the execution than in the assembly/establishment? Furthermore, the inheritance would depend on H3K9me3.

A GO classification based on the modified residue, and subnuclear localisation (Nuclear Lamina, perinucleolar) makes most sense to me, still. Colin

On Mon, Jan 6, 2020 at 3:39 PM Midori Harris notifications@github.com wrote:

yeah, we have nothing annotated to GO:0070734 or any descendant (including the MF GO:0046976), and no phenotypes mentioning H3K27

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/17160?email_source=notifications&email_token=ALZVLKDHH4RIMMQZRCA5I5LQ4M7DDA5CNFSM4HGIPE2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIFT2KA#issuecomment-571161896, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALZVLKF6TNIM5TUXOIBHFIDQ4M7DDANCNFSM4HGIPE2A .

[colinlog]

Yes this would be a better way, and if some care is taken to use the histone residues as the distinguishing feature for the different types of heterochromatin, the Pombe/fungal/invertebrate heterochromatin forms can be branches of the GO that are similarly structured to those of heterochromatin of mammals and plants. Another crucial kind of feature that could be leading or used in definitions are the qualifiers 'peri-centromeric' and 'sub-telomeric' heterochromatin. But there is quite some variation across the tree of eukaryotes on these chromosomal chromatin features, so that is somewhat idiosynchratic. The involvement of H3K9, which is a residue that becomes acetylated during transcription is the common theme. H3K9 methylation inhibits the advent of H3K9 acetylation and therefore locks the chromatin in a state that is less conducive to transcription. Conversely, when H3K9 or H3K27 are acetylated, they cannot be methylated without the intervention of a histone deacetylase that removes the acetyl group thus allowing the lysine to be methylated by a methyl-transferase. For organisms with H3K27 methylation the logic is the same. In fact H3K9and H3K27 both have the sequence context ARKS.

On Mon, Jan 6, 2020 at 3:51 PM Val Wood notifications@github.com wrote:

At the recent fission yeast meeting a chromatin expert (Rob Martienssen) also thought that "silencing" terms were a bit "outdated" , and that we could move to more "mechanistically" described terms

We describe by phenotype "silencing" mechanism "heterochromatin assembly" region pathway e.g. methylation dependent OR by small RNA (problematic because often both are involved)

and other ways...

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/17160?email_source=notifications&email_token=ALZVLKDEC5CWQK3G7DD2C2LQ4NAN5A5CNFSM4HGIPE2KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIFU6DI#issuecomment-571166477, or unsubscribe https://github.com/notifications/unsubscribe-auth/ALZVLKHAVS7PD7PDS3A72TDQ4NAN5ANCNFSM4HGIPE2A .

[ValWood]

we can close this since we will no longer have the "chromatin silencing" term it is out of date

[pgaudet]

This one is still there - we need to review the annotations before obsoleting.

[ValWood]

Ah right I thought all the silencing stuff was in a later ticket. This predates the obsoletion discussion.

[pgaudet]

Remaining terms in that branch:

chromatin silencing children (is_a, part_of and regulates)

chromatin silencing at telomere regulation of chromatin silencing regulation of chromatin silencing at telomere negative regulation of chromatin silencing negative regulation of chromatin silencing at telomere positive regulation of chromatin silencing positive regulation of chromatin silencing at telomere conversion of ds siRNA to ss siRNA involved in chromatin silencing by small RNA siRNA loading onto RISC involved in chromatin silencing by small RNA heterochromatin organization involved in chromatin silencing production of siRNA involved in chromatin silencing by small RNA

Are we obsoleting all ? Can we merge? (Need to look at annotations).

[ValWood]

We decided the chromatin silencing branch was not but we did not yet discuss:

conversion of ds siRNA to ss siRNA involved in chromatin silencing by small RNA siRNA loading onto RISC involved in chromatin silencing by small RNA production of siRNA involved in chromatin silencing by small RNA

chromatin silencing ones would be, for example GO:0031048 | heterochromatin assembly by small RNA

However, where compound term represent activities. So we have things like: production of siRNA involved in chromatin silencing by small RNA production of siRNA involved in RNA interference (so the loading/conversion/production terms are replicated in other branches)

So we would need to look at annotations and figure what the actual activities should be

the 'conversion conversion of ds siRNA to ss siRNA' has only 5 annotation (to argonaute and dicer).

In summary, this is a long-winded way of saying, yes these term are not required, but we did not discuss them yet....

[RLovering]

Hold it, I thought we were just looking at nuclear gene silencing, some of these terms are occuring at translational level eg: GO:0036404 conversion of ds siRNA to ss siRNA is a child of RNA processing!

We need to make sure that the role of dicer and argonaute is not lost through deletion of all 'silencing' terms.

I don't know anything about chromatin silencing to be able to comment on this, I do know that some of the miRs we have curated have their action at the DNA level and transcription rather than on mRNAs.

Can we wait a bit before deleting all the terms that describe the role of RNAs in gene silencing Thanks Ruth

[pgaudet]

Done in other tickets