merge 3706 & 4879? receptor txn factor activity terms

[gocentral]

David and I started to rearrange these two terms:

ligand-regulated txn factor activity (GO:0003706) (is_a "specific RNA polymerase II transcription factor activity") Def: Functions to enable the transcription of specific, or specific sets, of genes by RNA polymerase II in response to a ligand.

ligand dependent nuclear receptor activity (GO:0004879) (is_a "receptor activity") Def: A ligand-dependent receptor found in the nucleus of the cell.

We are unable to distinguish the two terms and have some questions that would be great to get input from the signalling group.

1. Is there any difference between 3706 and 4879?

David was thinking to add "receptor activity" as a parent for 3706, and then move 4879 as a child of 3706, like this:

- receptor activity (GO:0004872) -- ligand-regulated txn factor activity (GO:0003706) --- ligand dependent nuclear receptor activity (GO:0004879)

Trying to rename and redefine 3706

3706 would still have a parent under the transcription factor activity term, though I'll need to move it under a different term since the current parent will be going away. So, I was thinking to make it an is_a child of "sequence-specific DNA binding RNA polymerase II transcription factor activity" (GO:0000981) since the definition already specifies that the term is for RNAP II.

Then based on the two parents "receptor activity" (GO:0004872) and "sequence-specific DNA binding RNA polymerase II transcription factor activity" (GO:0000981), I was thinking:

- rename to "ligand-regulated sequence-specific DNA binding RNA polymerase II transcription factor activity"

- new def based on the defs of its new parent terms: Combining with an extracellular or intracellular messenger and interacting selectively and non-covalently with a specific DNA sequence in order to modulate transcription by RNA polymerase II.

Then I tried to rename and redefine 4879. For starters, I am not inclined to keep the nuclear localization part of the current def for 4879 since is where it is not how it acts. But neither David or I can figure out anything that distinguishes 4879 from the definition I wrote above for 3706. Any ideas?

2. The current 3706 term is already specific to RNAP II and the 4879 term requires nuclear localization. Is there any need to have a more general term, i.e. do prokaryotes have anything similar to CUP2 in S. cerevisiae, where it must bind copper in order to bind DNA.

3. Both of the current 3706 and 4879 terms are very general in their names, i.e. "ligand" rather than something more specific like "hormone". Thus as currently defined, the annotation of cerevisiae CUP2 to 3706 is appropriate. Do we need subtypes to distinguish what type of ligand, e.g. hormone, copper, etc.?

4. Can we come up with a definition for the function of this activity?

Combining the existing defs of "receptor activity" (GO:0004872) and "sequence-specific DNA binding RNA polymerase II transcription factor activity" (GO:0000981) can give this:

Combining with an extracellular or intracellular messenger and interacting selectively and non-covalently with a specific DNA sequence in order to modulate transcription by RNA polymerase II.

But David had some concerns with this: "I think the intracellular bit of this definition may be misleading. Any TF that is downstream of a signaling cascade would fit this (Notch-ICD for example). Classically these interact directly with the signaling molecules themselves." Then I wondered if copper could be considered a messenger, and thinking further, for the copper one, I'm pretty sure the messenger, if you can call it that, would have to be intracellular.

So, we felt that we needed to coordinate with the signalling group on these terms.

thanks,

-Karen and David

Reported by: krchristie

Original Ticket: geneontology/ontology-requests/7975

[gocentral]

Existing annotation info (11-23-2010):

GO:0003706 - ligand-regulated transcription factor activity

• 33 gene product associations

MGI ISO 2 RGD IDA 1 RGD IEA 1 RGD IMP 1 RGD ISO 4 RGD ISS 2 SGD IMP 1 SGD ISS 1 UniProtKB IDA 4 UniProtKB IEA 18 UniProtKB ISS 12 UniProtKB TAS 3

GO:0004879 - ligand-dependent nuclear receptor activity

• 779 gene product associations

Ensembl IEA 1 FB IDA 3 FB IMP 1 FB ISS 20 FB NAS 38 FB TAS 3 MGI IDA 8 MGI IEA 31 MGI IMP 1 MGI ISO 5 MGI NAS 1 MGI TAS 3 PDB IEA 46 RefSeq IEA 3 RGD IDA 4 RGD IMP 2 RGD ISO 14 RGD ISS 1 RGD NAS 1 RGD TAS 3 UniProtKB IDA 7 UniProtKB IEA 2668 UniProtKB ISS 17 UniProtKB NAS 2 UniProtKB TAS 21 WB IEA 41 WB ISS 1 ZFIN IDA 9 ZFIN IEA 98 ZFIN ISS 1

Scanning through the spreadsheets, the majority of annotations are clearly nuclear hormone receptors (the vast majority from UniProtKB by IEA).

So, if we merge, I think there's a clear preference to keep the 4879 term, since it has vastly more annotations.

Original comment by: krchristie

[gocentral]

Re (2), DtxR is a repressor rather than an activator, but apparently requires iron to bind DNA:

http://www.rose.brandeis.edu/PRLab/projects/transcription/transcription.html

But I agree with point (1) that nuclear localization isn't really a good criterion for distinguishing function terms anyway.

I kind of suspect that people wouldn't think of DtxR as a receptor, but that doesn't necessarily mean that its actions wouldn't fit a definition that we might use for a merged term. Maybe another angle on the merge question: is there any difference between requiring a bound ligand (ion, hormone, or whatever) to regulate transcription and being a receptor for the bound entity? If so, that would correspond to the distinction between 3706 and 4879; if not, that would argue in favor of merging.

m

Original comment by: mah11

[gocentral]

I am more-than-happy to see the 'nuclear' bit of GO:0004879 removed (I'd like to remove 'transmembrane' from the cell surface receptors too). Plus, some 'nuclear receptors' are found in the cytoplasm and move to the nucleus once they bind the ligand.

The main difference I can see ties in with David's point: the hormones originate from outside of the cell, whereras the copper (and other messengers) are intracelluar. So can we have two child terms?

ligand-activated transcription factor activity --%extracellular ligand-activated TF activity (is_a receptor) --%intracellular ligand-activated TF activity

extracellular ligand-activated TF activity (synonym: nuclear hormone receptor)

We can make a process-function link to 'intracellular receptor mediated signaling pathway ; GO:0030522'.

WRT Notch, PMID:12651094 is a nice review, and suggests that the NICD doesn't bind DNA itself, but instead binds to a DNA-binding TF, so wouldn't be eligible for any DNA-binding term.

There's still some debate about what a receptor is (and should things like ligand-gated ion channels (eg the P2X receptor) be given a receptor parent). I think to be a receptor, it has to bind/receive a primary signal. If copper is a secondary messenger, I'm not convinced CUP2 counts as a receptor.

thanks, Becky

Original comment by: rebeccafoulger

[gocentral]

Responding to Midori's comment about DtxR, that it "is a repressor rather than an activator, but apparently requires iron to bind DNA", I just wanted to mention that I've been using the phrase "transcription factor" in a very general way, such that it is appropriate for either activators or repressors, and then having child terms that are specific for either activation or repression. So, if we feel that there are both activators and repressors in this class of ligand-activated txn factors, I think we'd want to make child terms that specify each of those effects on transcription.

-K

Original comment by: krchristie

[gocentral]

I think Becky's question about the possibility of this type of structure:

ligand-activated transcription factor activity --%extracellular ligand-activated TF activity (is_a receptor) --%intracellular ligand-activated TF activity

is a really interesting idea.

Certainly when David and I were discussing this, I was wondering if CUP2 could be considered a "receptor" when it binds to copper, presumably that has already been internalized as I didn't seen any suggestion that CUP2 has any portion that is extracellular in order to pick up a signal there, but instead that the "CUP2 protein contains a cysteine-rich DNA-binding domain dependent on Cu+ and Ag+ ions which bind the cysteine residues and direct the refolding of the metal-free apoprotein." It does appear that CUP2 is effectively the copper sensor that detects when copper is present and needs to be responded to. I think there will be other similar situations, because at the transcription meeting I attended in June, someone spoke about a transcription factor that binds GTP, if I recall correctly.

Since it seems the main distinction between things like CUP2 and a typical nuclear hormone receptor is what type of signal the respond to, I think the signalling group might be in a better position to make a determination of whether what happens with CUP2 is a function that needs to be represented separately than the hormone receptor transcription factor receptors. But I don't have any problems with that idea from the transcription side, at least not that I can think of now.

-K

Original comment by: krchristie

[gocentral]

Hi Becky,

I used to work on RA receptors. I think that although people have called them nuclear receptors for years, that is a historical artifact about how they were first characterized. They do not always reside in the nucleus. Recent data shows that they can also repress or activate transcription depending on their partners and whether they have ligand bound. I think we should ditch the 'nuclear' terms and use that string as a synonym for a term that describes their actual function. I'll be interested in hearing what Ruth's colleague has to say.

D

Original comment by: ukemi

[gocentral]

Ruth's colleague Ines says:

1/ Do all nuclear hormone receptors/ligand-dependent nuclear receptors regulate transcription? Yes, as far as I know

2/ Do all nuclear hormone receptors/ligand-dependent nuclear receptors regulate transcription by binding DNA directly? Some nuclear receptors like GR (glucocorticoid receptor) can regulate transcription by either binding directly to DNA or indirectly via protein-protein interactions with another transcription factor (these sites are termed "tethering sites") This is not exclusive to GR but hasn't been described for all nuclear receptors

3/ Can you think of ligand-regulated transcription factors that aren't receptors? Ligand regulated is quite a broad term. I guess the issue is more ligand binding...I don't know of any Ts that are bound by a ligand that are not nuclear receptors but actually the transcriptional activity of most TFs are regulated by a compound or ligand for a membrane receptor (their phosphorylation, translocation etc which affects their activity)

4/ Is there a difference between the first two terms? I don't like the second term because it implies that all nuclear receptors are located in the nucleus which they are not. They act in the nucleus but they can be express in the cytoplasma and then upon hormone binding they translocate to the nucleus. GR and the androgen receptor behave that way. They belong to the subclass of steroid hormone receptors. The big family is nuclear receptors .I don't think nuclear hormone receptors is right to define the whole family since a lot of NRs (LXRs, PPARs, ROR, Reverb...) don't have hormones as ligands but still belong to the NR superfamily. The term nuclear hormone receptors is more classic (the family first started with the identification of the steroid hormone receptors) but doesn't reflect the heterogeneity of this class of TFs.

6/ Is CUP2 a receptor? (activates transcription in response to copper). Don't know this one. Does CUP2 bind to copper? Does the binding change CUP2 conformation so that other proteins/cofactors can bind it to promote transcription? You need at least proof of the first question to consider it a receptor I think...

Original comment by: rebeccafoulger

[gocentral]

I wanted to add some information about Cup2, which is currently annotated to the "ligand-regulated transcription factor activity" term. I don't know if Ruth could pass this on to her colleague to get her opinion, but it would certainly be great if that was possible.

Anyway, Cup2 definitely binds copper. The paper cited below shows that, and also shows that Cup2 only binds DNA when it is also binding copper. The authors say that binding copper causes restructuring of the copper binding domain such that it becomes a DNA binding protein. In the apo-metal (Cu or Ag) state, it is not capable of binding DNA. When copper is present intracellularly, Cup2 binds copper, and then binds its target site to activate transcription of metallothionein genes.

However, the authors refer to Cup2 as "the primary sensor of intracellular Cu+ in the yeast Saccharomyces cerevisiae" and Cup2 does not have transmembrane domains so it is clearly not functioning in exactly the same way that at least some of the hormone receptors do. So the question David and I had was whether or not to consider Cup2 as a receptor; we weren't sure.

Below is the title, abstract, and PMID of the paper which shows copper dependent DNA binding.

thanks,

-Karne

Buchman C, Skroch P, Welch J, Fogel S, Karin M (1989) The CUP2 gene product, regulator of yeast metallothionein expression, is a copper-activated DNA-binding protein. Mol Cell Biol 9(9):4091-5

Abstract: CUP2 is a regulatory gene controlling expression of CUP1, which encodes the Cu-binding yeast metallothionein. CUP2, which is identical to the ACE1 gene, encodes a Cu-regulated DNA-binding protein. The CUP2 protein contains a cysteine-rich DNA-binding domain dependent on Cu+ and Ag+ ions which bind the cysteine residues and direct the refolding of the metal-free apoprotein. CUP2 mutant alleles from Cu-sensitive yeast strains have point mutations affecting the DNA-binding activity. These results establish CUP2 as the primary sensor of intracellular Cu+ in the yeast Saccharomyces cerevisiae, functioning as a Cu+-regulated transcriptional activator. PubMed ID: 2674688

Original comment by: krchristie

[gocentral]

This existing term does reflect the new split between "nucleic acid binding" and "protein binding" transcription factors. Thus for the interim, Karen will probably make it a direct child of "molecular_function" in order to keep in in the ontology until the signalling group has had a chance to work through the issues.

In the long term, we may need to obsolete this term and replace it with a more specific term(s) that fits into the new structure of TF's in MF now.

-Karen

Original comment by: krchristie

[gocentral]

• labels: 397373 --> signaling
• assigned_to: nobody --> rfoulger

Original comment by: paolaroncaglia

[gocentral]

Reading the papers and people's comments, it sounds like Cu2 isn't a signaling receptor. Therefore I suggest the following edits, which would limit the use of ligand-dependent TF terms to proteins that are part of a signaling pathway:

1. Obsolete ligand-regulated transcription factor activity ; GO:0003706. -Annotations can be transferred to either the GO:0004879 terms or TF terms. -The Notch annotation can't be transferred because I don't think Notch binds DNA directly in the nucleus, but rather acts as a co-activator for other TFs. The Notch annotation could be moved up to 'signaling receptor activity'. -Reason for obsoletion: term is ambiguous.

2. Rename and redefine: ligand-dependent nuclear receptor activity ; GO:0004879 NEW NAME: ligand-activated transcription factor activity ; GO:0004879 Def: Combining with a signal and transmitting the signal to the transcriptional machinery by interacting selectively and non-covalently with a specific DNA sequence in order to modulate transcription by RNA polymerase II. is_a: signaling receptor activity ; GO:0038023 is_a: sequence-specific DNA binding RNA polymerase II transcription factor activity ; GO:0000981 part_of: intracellular receptor mediated signaling pathway ; GO:0030522 (P/F link) narrow synonym: nuclear hormone receptor related synonym: ligand-dependent nuclear receptor activity

3. Create new term for annotation of CUP2:

   • copper ion-activated DNA-binding transcription factor activity is_a: sequence-specific DNA binding RNA polymerase II transcription factor activity ; GO:0000981 part_of: cellular response to copper ion ; GO:0071280 (P/F link) HAS_PART: copper ion binding ; GO:0005507 Def: Interacting selectively and non-covalently with a specific DNA sequence in order to modulate transcription by RNA polymerase II, when bound to copper (Cu) ions. PMID: 2674688

I'd like to commit any edits before the upcoming GO meeting. Any comments/concerns?

thanks Becky

Original comment by: rebeccafoulger

[gocentral]

Hi Becky,

Regarding #1, I have no objections to obsoleting "ligand-regulated transcription factor activity ; GO:0003706" for the reason you've stated.

Regarding #'s 2 and 3, I have some comments. For #3, I would like to talk to David about the basic idea of encoding MF terms such as the one you've proposed. But I'll go ahead and make a couple of comments now.

2. Since you've given a definition that is specific to RNAP II and requires sequence specific DNA binding, please also make the term name specific to those two details. A term named "ligand-activated transcription factor activity" would need to be broadly applicable to multiple RNAPs. So, for the term name to be consistent with its position in the structure and existing term names, I think it would need to be this:

ligand-activated sequence-specific DNA binding RNA polymerase II transcription factor activity

3. Pending discussion with David about encoding terms with ion binding requirements, I am still considering if this is a direction we should go.

If we do create terms like this, I have the same comment about the proposed term name for this one as above. The proposed term name would need to be for a term more broadly applicable than the proposed definition. Considering that the DtxR gene that Midori mentioned in this thread is bacterial, it seems that we might need both the general and the RNAP II specific instances for this term.

In any case, as a general rule, I have tried to make the term names specific when the definitions are to avoid misannotation by people not reading the def to realize it is more specific than one might assume from the name.

thanks,

-Karen

Original comment by: krchristie

[gocentral]

This is now done.

Original comment by: rebeccafoulger

[gocentral]

• status: open --> closed-fixed

Original comment by: rebeccafoulger