#9895 issues with def of "NURF complex"

MGI
closed-accepted
5
2013-02-22
2012-12-11
Karen Christie
No

Hi,

The def for the CC term "NURF complex" looks like it got written a long time ago. I think it needs to be updated. Here is the current:

A four subunit ISWI-containing protein complex that facilitates nucleosome mobility and transcriptional activation in an ATP-dependent manner. In contrast to other chromatin remodeling complexes, the ATPase activity of NURF requires nucleosomes rather than free DNA or histones.

However, this review:

Alkhatib SG, Landry JW. The nucleosome remodeling factor. FEBS Lett. 2011 Oct
20;585(20):3197-207. doi: 10.1016/j.febslet.2011.09.003. Epub 2011 Sep 9. Review.
PubMed PMID: 21920360.

indicates that this def is specific to the Drosophila form of the complex, while the mammalian form is slightly different:

NURF was first identified in Drosophila melanogaster as an ATP-dependent biochemical activity that enhanced GAGA factor (GAGAG binding factor)-mediated nuclease accessibility to reconstituted chromatin [5] and [6]. Biochemical purification of this activity revealed a four subunit complex composed of NURF301, the ATPase ISWI (NURF140), NURF55 and NURF38 polypeptides [7]. Purifications from human cells identified a complex highly homologous to D. melanogaster NURF, strongly suggesting that it has been conserved through evolution [8]. Homo sapiens NURF contains the NURF301 homolog BPTF, the ISWI homolog SNF2L, and a NURF55 homolog pRBAP46/48; however, a NURF38 homolog has not been identified [8]

Here is a proposed new def:

Def: An ISWI-type chromatin remodeling complex that facilitates nucleosome mobility in an ATP-dependent manner. In contrast to some other chromatin remodeling complexes, the ATPase activity of NURF requires nucleosomes rather than free DNA or histones. In Drosophila, where the complex was first identified, it consists of four subunits, NURF301, the ATPase ISWI (NURF140), NURF55 and NURF38. In humans, this complex contains BPTF (a NURF301 homolog), SNF2L (an ISWI homolog), and pRBAP46/48 (a NURF55 homolog), but a NURF38 homolog has not been identified. Note that mammals have two ISWI homologs, SNF2L and SNF2H; NURF forms preferentially with SNF2L. Note also that alternately spliced forms of some of these genes have been observed in multiple organisms, so there may be some variability in the exact protein isoforms present in NURF family complexes.

Note: The phrase "nucleosome remodeling factor complex" is currently an exact synonym of "NURF complex", and the NURF name is based on this phrase. However, this phrase is equally true for CERF as it is also a "nucleosome remodeling factor complex". Not sure how best to deal with this, but wanted to point it out.

thanks,

-Karen

Discussion

1 2 > >> (Page 1 of 2)
  • My only suggestion is to move this part:

    "In Drosophila, where the complex was first identified, it consists of four subunits, NURF301, the ATPase ISWI (NURF140), NURF55 and NURF38. In humans, this complex contains BPTF (a NURF301 homolog), SNF2L (an ISWI homolog), and pRBAP46/48 (a NURF55 homolog), but a NURF38 homolog has not been identified. Note that mammals have two ISWI homologs, SNF2L and SNF2H; NURF forms preferentially with SNF2L. Note also that alternately spliced forms of some of these genes have been observed in multiple organisms, so there may be some variability in the exact protein isoforms present in NURF family complexes."

    into the comment. Other than that, the update sounds good.

     
    • assigned_to: nobody --> tairtb
     
  • Karen Christie
    Karen Christie
    2012-12-12

    That's fine with me Tanya, whatever is best practice now.

    That will leave the definition almost unchanged to what is there now. Removing the part about 4 subunits is required. I also removed the part about transcription activation because it appears that these chromatin remodelers affect many DNA based processes, so it seemed better to just keep the def specific to the direct effect.

    -Karen

     
  • Karen Christie
    Karen Christie
    2012-12-12

    Actually, I'm having second thoughts about moving the subunit info out of the def. The only thing that I know of that distinguishes NURF from CERF is the subunit composition.

    How should we deal with that?

    -Karen

     
  • Good point. How about this: (merging the two issues now so we can see them side by side)

    CERF:Def: An ISWI-type chromatin remodeling complex that facilitates nucleosome mobility in an ATP-dependent manner. In contrast to some other chromatin remodeling complexes, the ATPase activity of CERF requires nucleosomes rather than free DNA or histones. In humans, this complex contains at least SNF2L and CECR2.

    NURF:Def: An ISWI-type chromatin remodeling complex that facilitates nucleosome mobility in an ATP-dependent manner. In contrast to some other chromatin remodeling complexes, the ATPase activity of NURF requires nucleosomes rather than free DNA or histones. In humans, this complex does not contain CECR2.

    Hmmmmmmm. Now I'm thinking that this is really something for PRO since the distinctions are based on subunit composition.

    Maybe what we really need is a more generic term for both CERF and NURF, say 'nucleosome remodeling factor complex' with narrow synonyms of CERF and NURF and the def that is the same between the two proposed above:

    An ISWI-type chromatin remodeling complex that facilitates nucleosome mobility in an ATP-dependent manner. In contrast to some other chromatin remodeling complexes, the ATPase activity of the complex requires nucleosomes rather than free DNA or histones.

    What do you think of that?

     
  • Karen Christie
    Karen Christie
    2012-12-12

    Yea, we probably do need to take a step back from this. I was trying to avoid looking at a term I don't need (the "ISW1 complex" term), but I now think we should consider two more terms also, the parent term of "NURF complex" and it's existing sibling "ISW1 complex". This is the basic structure of the terms to consider:

    GO:0031010 - ISWI complex
    is_a GO:0016587 - ISW1 complex
    is_a GO:0016589 - NURF complex
    is_a GO:proposed - CERF complex

    What is the difference between the ISW1 complex term and its parent term "ISWI complex"

    ISWI complex - Any chromatin remodeling complex that contains an ATPase subunit of the ISWI family.

    ISW1 complex - A protein complex that contains an ISWI-family ATPase such as Saccharomyces Isw1p, and acts to modify chromatin structure.

    While the ISW1 complex def does list a specific example, it doesn't really seem to restrict the definition beyond the def of its parent term for "ISWI complex", so I am wondering the purpose of the "ISW1 complex" term.

    At this point, it might be good to lood at a general review of ISWI type complexes to see if there are some generally recognized types.

    I have also asked Harold to take a look at this item. I had already asked him for PRO advice on the CERF paper anyway.

    -Karen

     
  • Karen Christie
    Karen Christie
    2012-12-12

    Here are some reviews, along with some comments from a quick glance at them.

    1: Lange M, Demajo S, Jain P, Di Croce L. Combinatorial assembly and function of
    chromatin regulatory complexes. Epigenomics. 2011 Oct;3(5):567-80. doi:
    10.2217/epi.11.83. Review. PubMed PMID: 22126247.

    - still waiting to get full text of this one

    2: Alkhatib SG, Landry JW. The nucleosome remodeling factor. FEBS Lett. 2011 Oct
    20;585(20):3197-207. doi: 10.1016/j.febslet.2011.09.003. Epub 2011 Sep 9. Review.
    PubMed PMID: 21920360.

    - specific to NURF, alludes to possibility of multiple compositions

    3: Erdel F, Rippe K. Chromatin remodelling in mammalian cells by ISWI-type
    complexes--where, when and why? FEBS J. 2011 Oct;278(19):3608-18. doi:
    10.1111/j.1742-4658.2011.08282.x. Epub 2011 Sep 2. Review. PubMed PMID: 21810179.

    - has awesome diagram showing various classes of ISWI complexes and compositions. I think this is a great candidate to consider for making GO terms for ISWI classes of complexes

    4: Racki LR, Narlikar GJ. ATP-dependent chromatin remodeling enzymes: two heads
    are not better, just different. Curr Opin Genet Dev. 2008 Apr;18(2):137-44. doi:
    10.1016/j.gde.2008.01.007. Epub 2008 Mar 12. Review. PubMed PMID: 18339542;
    PubMed Central PMCID: PMC2494867.

    - compares ISWI to SWI/SNF class

     
  • Is there a difference in the processes facilitated by CERF vs. NURF? If so, let's include that in the definitions and make the new term for CERF. Though there are very many protein complex terms in GO that are only defined by subunit composition, I do feel that we should only add new protein complexes of very similar composition when they can be associated with different processes/functions. PRO did not exist when many of the 'subunit-y' complexes were added to the GO which may explain those terms' existence. However, now that there is PRO, I think we should try to set limits on what we add to GO and what can better be represented by PRO.

     
  • Karen Christie
    Karen Christie
    2012-12-12

    HI Tanya,

    Yes, there are differences in teh processes for CERF vs NURF. Funnily enough, I already had a similar discussion with David with respect to whether or not to even ask for the CERF complex term!

    The Erdil & Rippe review I mentioned in a previous comment has a really awesome diagram showing the compositions and overall processes of the mammalian ISWI complexes. I think it would be a good basis for making several child ISWI complex terms. However, the title of the review that I have on request talks about combinatorial assembly, so I think I'd like to wait a few days to get that before proposing specific terms based just on the Erdil & Rippe review, if you don't mind.

    -Karen

     
  • Great minds think alike. :) I hope the review is useful. I just checked if Stanford gets that journal and I can't get to it either. I'll stay tuned.

     
  • Karen Christie
    Karen Christie
    2013-01-14

    Hi Tanya,

    This whole area looks like a mess that hasn't been looked at in a while. Amongst other
    things, we have two classification schemes being used simultaneously, one based on
    composition and a second based on function, e.g. "ISWI complex" and "chromatin assembly
    complex". These are overlapping. The parts of the "function" based classification that
    I've looked at so far seem largely based on outdated thinking.

    The Lange et al. 2011 review looks like it has good info on several ATP-dependent chromatin
    remodeling families (BAF, CHD, INO80, and ISWI). However, I only have time to poke into
    the ISWI family, since that is where I need my term, so these comments are mostly limited
    to that, except for the fact since I want to move a term from "chromatin assembly complex"
    to "chromatin remodeling complex", I have come across a higher level issue. For this, I
    have a proposal that I think is an improvement, though not a complete resolution of the
    issues in this area.

    Here is the summary of affected existing and new terms, and below a summary of the
    structural changes I propose. I will put the specifics of proposed name and definition
    changes into the text file I will attach; hopefully that will be easier than copy/pasting out of the SF item. I can also send you text file directly if you prefer.

    thanks,

    -Karen

    Existing terms with proposed modifications:
    --------------------------------------------
    GO:0031010 - ISWI complex
    GO:0016587 - Isw1 complex
    GO:0016589 - NURF complex
    GO:0031213 - RSF complex
    GO:0016590 - ACF complex
    GO:0008623 - chromatin accessibility complex (aka CHRAC)
    GO:0005678 - chromatin assembly complex
    GO:0016585 - chromatin remodeling complex

    Proposed new complex terms
    --------------------------------------------
    GO:new1 - WICH complex
    GO:new2 - NoRC complex
    GO:new3 - CERF complex
    GO:new4 - ATP-dependent chromatin remodeling complex

    These two diagrams show the proposed structure changes. I am just showing affected terms
    (via is_a links) in this area; child terms of "chromatin remodeling complex" that are not affected have been left out.

    current structure:
    --------------------------------------------
    - chromatin assembly complex - GO:0005678
    -- ACF complex - GO:0016590
    -- CAF-1 complex - GO:0033186
    -- HIR complex - GO:0000417
    - chromatin remodeling complex - GO:0016585
    -- INO80-type complex - GO:0097346
    -- ISWI complex - GO:0031010
    --- Isw1 complex - GO:0016587
    --- NURF complex - GO:0016589
    -- RSF complex - GO:0031213
    -- chromatin accessibility complex - GO:0008623
    -- histone deacetylase complex - GO:0000118
    --- NuRD complex - GO:0016581
    -- SWI/SNF-type complex - GO:0070603

    possible new structure (with new4)
    --------------------------------------------
    - chromatin assembly or remodeling complex - GO:0016585
    -- ATP-dependent chromatin remodeling complex - GO:new4
    --- INO80-type complex - GO:0097346
    --- ISWI complex - GO:0031010
    ---- Isw1 complex - GO:0016587
    ---- NURF complex - GO:0016589
    ---- RSF complex - GO:0031213
    ---- ACF complex - GO:0016590
    ---- CHRAC - GO:0008623
    ---- WICH complex - GO:new1
    ---- NoRC complex - GO:new2
    ---- CERF complex - GO:new3
    --- SWI/SNF-type complex - GO:0070603

    -- histone deacetylase complex - GO:0000118 (see text file for comment)
    --- NuRD complex - GO:0016581

    -- histone DNA-deposition complex - GO:0005678
    --- CAF-1 complex - GO:0033186
    --- HIR complex - GO:0000417

    Though, if you don't feel comfortable with moving the GO:0005678 term to be a child of
    GO:0016585, we could leave it as a sibling.

     
  • Karen Christie
    Karen Christie
    2013-01-14

     
    Attachments
  • Email discussion transcribed:

    I've been reading your proposal and thinking about possible solutions. I must say that I don't like the idea of renaming 'chromatin remodeling complex' as 'chromatin assembly or remodeling complex' and renaming 'chromatin assembly complex' as 'histone-DNA deposition factor'. There's too much process-y stuff going on.

    In fact, I'm leaning more towards obsoletion of the 'process' based complex terms and retaining only the composition-based terms. Curators can annotate to the composition-based term for complex, if they wish, and annotate to the actual process (chromatin assembly and/or chromatin remodeling as the case may be) to capture that information.

    That may very well leave us with a very flat structure but I think that is acceptable.

    -INO80-type complex
    -ISWI complex
    ----bunch of kids
    -SWI/SNF-type complex
    -histone deacetylase complex
    ----NuRD complex
    -CAF-1 complex
    -HIR complex

    obsolete:
    chromatin assembly complex
    chromatin remodeling complex

    Alternatively, if we don't want to obsolete those two terms for sentimental reasons, we could pop out the children that do both remodeling and assembly to be siblings of the above terms, under the protein complex parent.

    Example:

    -chromatin assembly complex
    -chromatin remodeling complex
    -IMATAR (incredible mega-assemblage that assembles and remodels) complex

    What do you think?

     
  • Karen's response
    -----------------------
    HI Tanya,

    I think I would prefer to just completely obsolete the two higher level problem terms:
    - chromatin assembly complex
    - chromatin remodeling complex

    It seems like it could be confusing to to keep them and then then have the terms for several complexes that get called a "chromatin remodeling complex", not under that term because they also do assembly.

    I do think it might be good to create a new term to group everything in the SWI2/SNF2 superfamily. That was the new4 term I already proposed, but we could add the name of the SWI2/SNF2 superfamily name to make it explicit that it is a grouping based on compositional similarity. I have not included assembly in the name everything I have seen just calls them "ATP-dependent chromatin remodelers", even though lots of them also do assembly. But maybe we could just make the definition broad enough to be clear that assembly is one of the activities found in this family.

    The intro of the Dirscherl & Krebs 2004 lists 7 or so subfamilies and cites a couple reviews that might help. The four groups I've included below are the four that the Lange et al. 2011 review lists. Then an existing term like "NuRD complex" can get additional new compositional parentage to group it appropriately within the SWI2/SNF2 superfamily based complexes. So, below are some modifications to what I proposed before with changes from my original idea in red.

    -- SWI2/SNF2 superfamily ATP-dependent chromatin remodeling complex - GO:new4
    --- INO80-type complex - GO:0097346
    --- ISWI-type complex - GO:0031010
    --- CHD-type complex – GO:new
    ---- NuRD complex - GO:0016581
    -- SWI/SNF-type complex - GO:0070603

    --- histone deacetylase complex - GO:0000118
    ---- NuRD complex - GO:0016581

    Would something like this work for you?

    -Karen

    P.S. Interestingly, two other child terms of the "histone deacetylase complex", the Rpd3L-Expanded complex and the Snt2C complex terms, also indicate that they are chromatin remodeling complexes as well as histone deacetylases, but I don't recognize the type of chromatin remodeling complex they are, so I can't propose any additional parentage based on the chromatin remodeling composition.

     
  • Hi Karen,

    In your last suggested structure, should SWI/SNF-type complex - GO:0070603 be a child of SWI2/SNF2 superfamily ATP-dependent chromatin remodeling complex -
    GO:new4 ? Also how are the last two terms related to the first set? From the display it looks like everything is a child of GO:new4 except for GP:0070603, which is a sibling.

    Thanks,

    Tanya

     
  • Hi Karen,

    In your last suggested structure, should SWI/SNF-type complex - GO:0070603 be a child of SWI2/SNF2 superfamily ATP-dependent chromatin remodeling complex -
    GO:new4 ? Also how are the last two terms related to the first set? From the display it looks like everything is a child of GO:new4 except for GP:0070603, which is a sibling.

    Thanks,

    Tanya

     
  • Karen Christie
    Karen Christie
    2013-01-16

    HI Tanya,

    Yes, sorry, my error. The existing "SWI/SNF-type complex - GO:0070603" would need to be a child of a term for "SWI2/SNF2 superfamily ATP-dependent chromatin remodeling
    complex" if it is created.

    thanks,

    -Karen

     
  • Sorry to pester but what about the second part of my last comment:

    "Also how are the last two terms related to the first set? From
    the display it looks like everything is a child of GO:new4 except for
    GP:0070603, which is a sibling."

    Thanks.

     
  • Karen Christie
    Karen Christie
    2013-01-17

    HI Tanya,

    Sorry I missed that part.

    This part is the existing parentage of the NuRD complex, so no change would be needed here, just showing the context that it's in now. But if we are going to implement a composition based structure for SWI2/SNF2 superfamily chromatin remodelers, it should get additional parentage.

    --- histone deacetylase complex - GO:0000118
    ---- NuRD complex - GO:0016581

    -Karen

     
  • After a review of the implications of obsoleting 'chromatin assembly complex' and 'chromatin remodeling complex', which were quite large, Karen and I came up with an alternate solution:

    Merge the terms instead:

    GO:0005678, chromatin assembly complex
    GO:0016585, chromatin remodeling complex

    resulting in GO:0005678, chromatin assembly or remodeling complex

    A new grouping term for the "SWI2/SNF2 superfamily chromatin remodeling complex" will also be created as a composition based grouping term for the subset of chromatin remodeling or assembly complexes that fit into that group. That would move us a step in the direction of a more compositionally based structure.

     
  • Hi Karen,

    The discussion goes on but I think it's gotten better! I brought up the issue during this morning's ontology development call and here's the suggestion that came out of it, courtesy of Chris.

    Create a NEW term called 'chromatin assembly and remodeling complex' that will cover the complexes that are involved in BOTH processes. The new basic structure will be like this:

    chromatin assembly complex
    ---[i] chromatin assembly and remodeling complex ; GO:new

    chromatin remodeling complex
    ---[i] chromatin assembly and remodeling complex ; GO:new

    We can then sort the composition based terms into the three process-based categories:

    assembly only
    remodeling only
    both

    Does that make sense? It seems to solve our situation pretty neatly. Please let me know if I've missed some important detail that makes this solution not viable.

    Thanks,

    Tanya

     
  • Karen Christie
    Karen Christie
    2013-01-24

    Hi Tanya,

    Perhaps I'm missing something, but I think I like this suggestion less well. From most of the reviews I've seen, people group these by what type of ATPase they have by family, e.g. ISWI, INO80, i.e. a compositional grouping. Then within the ISWI family, there are some that only do remodelling and some that do both remodelling and assembly. I think it would be confusing to split the ISWI family across two different process based classes.

    Let me know if I misunderstood something about Chris's suggestion.

    thanks,

    -Karen

     
  • Can't we have both? Two parallel sorting options, one by subunit composition and one by process.

    protein complex
    -- SWI2/SNF2 superfamily ATP-dependent chromatin remodeling complex -
    GO:new4
    ----INO80-type complex - GO:0097346
    ---- ISWI-type complex - GO:0031010
    ------ ISWI-type complex that only does remodeling
    ------ ISWI-type complex that does both
    ---- CHD-type complex – GO:new
    ------ NuRD complex - GO:0016581
    ---- SWI/SNF-type complex - GO:0070603

    chromatin assembly complex
    ---[i] chromatin assembly and remodeling complex ; GO:new

    chromatin remodeling complex
    ---[i] ISWI-type complex that only does remodeling
    ---[i] chromatin assembly and remodeling complex ; GO:new
    ------[i] ISWI-type complex that does both

     
  • Karen Christie
    Karen Christie
    2013-01-24

    At least based on the ISWI family, making the process based grouping to separate into remodeling only versus both remodeling and assembly seems a little artificial. It seems like this is just one of those cases where the original name, i.e. remodeling complex, was given based on only a partial understanding of the range of activities. To maintain these groupings based on the "assembly complex" and "remodeling complex" names just seems like it's propagating a narrow view that no longer exists for the SWI2/SNF2 superfamily. People still call them remodeling complexes, and then list assembly as one of the activities that a remodeling complex can do.

    Then if you look beyond these SWI2/SNF2 complexes to the other complexes currently under "chromatin assembly complex" and especially "chromatin remodeling complex", they are incredibly diverse having little or no relationship to each other, including histone acetyltransferase and histone deacetylase complexes, as well as ATP-dependent chromatin remodelers. Perhaps it would be better not to try to group complexes based on process within the Component ontology, and let the process annotations take care of grouping by process.

    my two cents

    -Karen

    P.S.

    It looks like the "nucleosomal methylation activator complex" (GO:0070311) is also a SWI2/SNF2 superfamily complex, of the BAF subtype.

    should be a type of "SWI/SNF-type complex" (GO:0070603), which in turn should have a synonym for "BAF-type complex". Due to its position as a sibling of "ISWI complex" and "INO80-type complex", and also the fact that most of its child terms appear to be in the BAF subfamily, I had been thinking that this term was the "BAF-type complex" term, but the def of "SWI/SNF-type complex" (GO:0070603) is broad enough to be the parent term for all of the SWI2/SNF2 superfamily terms.

    If the problem is a grouping term with an "or", e.g. "chromatin assembly or remodeling complex", I think I would consider getting rid of both "chromatin assembly complex" and "chromatin remodeling complex" completely. I would still add the proposed term to group the subclasses of the "SWI2/SNF2 superfamily ATP-dependent chromatin remodeling complex".

     
  • We are circling back to obsoletion. Perhaps we should send out the obsoletion proposal email and gauge the reaction.

     
1 2 > >> (Page 1 of 2)