Hi Val,
Responses in red:
Thanks for this it is very useful. Could you confirm a few things for me to check I
haven't misinterpreted. The situation is as I thought for pombe, but we are trying
to find a GO solution which would allow queries to retrieve the functionally
equivalent genes across species.
I guess you would have to call them elements, regions, features or something,
rather than genes.
1. If I have interpreted correctly, for organisms with 'regional ' centromeres
See below for expanded answers.
a) There is a distinct 'core region' in all organisms,
~yes
and the common defining feature of this region is that it is
b) not heterochromatic
~yes
c) enriched for CENP-A
Yes
d) Is the region of kinetochore attachment Region of kinetochore assembly and
microtubule attachment
Yes.
e) As you move away from the 'central core' region the amount of CENP-A
decreases and the DNA is assembled into heterochromatin (=pericentromeric
heterochromatin /centromeric heterochromatin)
~yes. This is the picture in e.g. human, Drosophila (but the term ‘central core’
not used)
--In pombe this region is not repetitive but in other organisms, including human,
it is
Yes. In humans the whole of the centromere is repetitive (CENP-A and
heterochromatin). In pombe the central core is ‘unique’ and the outer repeats are
repetitive (but far, far larger than human alpha satellite repeats).
Q Are a-e above correct ?
Mostly. Parts c and d are true (with the modification I wrote above). Part b is
correct as far as we know – let’s say it’s our understanding. But when DNA is
repetitive it can be pretty hard to assess – there could be mixing, but basically b
is correct. Some organisms don’t have centromeric heterochromatin at all (so,
whilst it’s true that that the CENP-A domain doesn’t have heterochromatin, it’s
also not relevant in those organisms).
Part a – it is true that there is a CENP-A domain in all organisms (except
Trypanosmes), but it wouldn’t be called a core region necessarily. ‘Central
region’ / ‘core region’ are sometimes found, but it’s by no means universal. See
below.
Q Does this 'central' region have any names in higher eukaryotes to distinguish
this region from the "centromeric/pericentric heterochromatin"?
Probably the most common phrase you would find is something along the lines of
‘central CENP-A-containing region’. Although many/most regional centromeres
have central CENP-A-containing region flanked by heterochromatin, this isn’t true
for all of them – for instance, Candida albicans centromeres have CENP-A
domains spanning a few kb but has no classical centromeric heterochromatin (it
lacks an H3K9-histone methyltransferase).
The main thing is that although the proteins found at centromeres are conserved
the DNA is completely different between species. For a pictorial representation
of DNA elements and chromatin domains see the Figure (and extensive figure
legend) in our review - attached. It’s a few years old now, but the story is
basically the same.
2. For GO terms would the following work for all species with modular
centromeres?
(BTW, Modular centromeres is not a generally used term. You are using it to
mean complex centromeres with distinct domains. Regional centromeres is an
often used (but it is perhaps not all that helpful as there is great variation
between regional centromeres of different species.)
If we defined :
i) A generic "centromere complex" term which included the specialised
chromatin with CENP-A assembled and the kinetochore proteins
ii) A 'sibling' term "centromeric/pericentric heterochromatin", and
iii) A broader 'parent' term "centromeric region" which included both the core
domain associated proteins of the centromere complex, and the heterochromatic
region?
This would allow people to query on "centromere complex" for the proteins of the
central chromatin kinetochore platform, and the kinetochore itself, pericentric
heterochromatin if they are only interested in the proteins at the
heterochromatic/pericentric regions or the more general "centromeric region" if
interested in all the proteins which are localised to both the kinetochore
associated region or the entire region associated with centromere sequence.
I’m no expert on GO terms. I think the basic structure of this tree (is it called a
tree?) is fine. See below for a few suggestions for the terms that would be most
appropriate. The term ‘centromere complex’ is not really used and might be
confusing – it might be expected to include DNA.
Centromeric proteins
 “Kinetochore Complex” and “Centromeric Chromatin”
 (Peri)centromeric Heterochromatin
Centromeric proteins
 Kinetochore Proteins and CENP-A Chromatin
 (Peri)centromeric Heterochromatin
Or some variation on these.
A few more things to take into consideration:
Are you planning on including histone marks/modifications in your definitions? I
guess you wouldn’t normally do so, but they are key features of centromeres.
CENP-A is easy – it’s a distinct protein. But the key feature of heterochromatin is
methylation (di and/or tri) on histone H3 lysine 9.
As I mentioned previously, there is some histone H3 in the central domain in
pombe – it’s not all CENP-A. This is the case in other organisms as well – there
is H3 within the CENP-A domain and there are a number of publications on the
composition of centromeric chromatin (ie the non-heterochromatic part)
especially in vertebrates & flies, the post-translational modifications of the
histones within it (especially H3), and the function (e.g. association of CENP-T
etc). I can supply you with these if you want them.
In addition, other histone variants may be present / absent within centromeric
regions. For instance H2Az is low in pombe central domain.
So, it’s all a bit complicated and may differ between organisms (or we don’t have
the full picture in all organisms).
We may think of the centromere in cartoon, linear form, but actually in cells the
centromeric DNA, heterochromatin and kinetochore forms a three-dimensional
structure. There are several papers on this topic. It also means that the term
‘outer’ for instance, might be used to mean the outer edges of a linear structure
(which would be expected to be heterochromatic) or it might be used to mean the
outer surface of the kinetochore than is CENP-A containing and contains
kinetochore proteins and connects to the microtubules. See Fig 2 of the Stellfox
review I’ve attached.
3. Note that we will use the sequence ontology to define the individual repeats
in fission yeast :
The children of "regional centromere" which is the defined as a sequence region:
http://sequenceontology.org/browser/current_svn/term/SO:0001795
and the children of this can be used to capture the pombe specific repeats dg/dh
etc
This sounds fine.
What we are doing in GO is different, here we are aiming to capture the
recognised cellular components (protein and protein-DNA complexes) rather than
the sequence features per. se.
We will be able to use both in annotations depending which is most appropriate
for the experiment.
I am also CCing GO collaborators David Hill (GO editor at MGI) and Anna
Melidoni (human GO curator at UCL) as we are trying to resolve some tricky
issues with the existing GO graph. They may have a couple more questions to
make the solution work for higher eukaryotes.
Many thanks, this was really helpful,
Val
On 03/12/2013 10:57, Alison L Pidoux wrote:
Hi Val,
Apologies for my delay in replying.
Which precise regions (with relation to teh repeats) are people referring to when
they use the phrases
1. Centromeric heterochromatin
2. Pericentric heterochromatin
(are these 2 terms interchangeable or is there a subtle difference?
They're talking about the same thing. So, historically we said 'centromeric
heterochromatin' in pombe, or 'outer repeat heterochromtin'. 'Pericentromeric
heterochromatin' was used to describe the heterochromatin at/adjacent to
centromeres in mammalian cells etc. However, it began to be used in pombe (I
believe partially to emphasise the similarity between the centromeres of pombe
and multicellular eukaryotes). So, it's the same thing, they are used
interchangeably.
3 is the "centromere core region" heterochromatic?, I have seen it described as
specialised chromatin, but not as heterochromtin.
The central core region in NOT heterochromatic. The central core is packaged in
CENP-A chromatin, which is entirely different from euchromatin and
heterochromatin. The main defining feature is that in central core region (or
central domain, as Robin prefers) is that (most) histone H3 is replaced by the
histone H3 variant CENP-A. The CENP-A chromatin forms the platform upon
which the kinetochore is assembled. This is the case in all eukaryotes (except
Trypanosomes) - CENP-A chromatin is assembled at centromeres and the
kinetochore is built upon it (in other organisms it's not called the central
core/domain/region).
4. When people refer to the "centromere" is the "pericentric heterochromatin"
include in the definition (I guess this is partially dependent on the answers to 1 &
2?
What I mean by centromere in pombe is the sequence encompassing the central
domain plus the outer repeats (ie the whole ~40 kb for cen1), with the CENP-A
chromatin and kinetochore plus the heterochromatin on the outer repeats. So,
yes, it would include the pericentromeric heterochromatin. I think that most
people would use the term in this way for pombe.
5. Are there any species differences in the use of these terms ?
Yes, I'd say it's variable. This is partly historical and partly due to the fact that
centromeric sequences are not conserved between species, and the set-up of
centromeres varies between species. In human, for instance, centromeric DNA
is composed of megabases of alpha-satellite (171 bp repeat). It appears that the
'good' repeats are towards the centre and have CENP-A and as you move
outwards the repeats are degraded and are assembled in heterochromatin
(=pericentromeric heterochromatin). There isn't the defined domain structure
that you find in pombe. Mammalian centromeres are far harder to analyse due to
their highly repetitive nature.
Here is a general description of pombe centromeres:
The central domain is composed of the central core and the inner part of the
innermost repeats (imr). The central domain is assembled in CENP-A chromatin
in which canonical H3 is replaced with CENP-A. The kinetochore (proteins such
as Mis6, Cnp3 etc) is assembled on the CENP-A chromatin, and microtubles
attach to the kinetochore.
There is some H3 in the central domain (estimated to be around 10% of what you
find elsewhere in the genome - but don't quote me on that). The function of the
H3 in the central core is not known (but H3 is found at mammalian and
Drosophila centromeres also, along with CENP-A). It is thought that CENP-A is
in an octomeric nuclesome that also contains two copies of H4, H2A and H2B
(this is a hugely controversial subject in centromere biology at present).
tRNAs are found in the imr repeats. See diagrams in attached ppt. The outer
tRNAs of imr1 have been shown to have boundary function (Kristin Scott). So
they form a boundary between the CENP-A/kinetochore domain and the
heterochromatin domain of the outer repeats.
There are also clusters of tRNAs at 5 of the 6 centromere extremities, suggesting
a boundary function to prevent the spread of heterochromatin from the outer
repeats to the euchromatin. There are also IRC elements at the edges of
centromeres which have boundary function (Grewal etc).
The outer repeats (dg/dh or K, depending on if you use Yanagida's or Clarke's
nomenclature) are assembled in RNAi-directed heterochromatin. Histones are
underacetylated and methylated on Histone H3 lysine 9 (dependent on Clr4).
Functions of the domains:
Central domain: assembly of CENP-A chromatin. Site of kinetochore assembly.
Outer repeats: Heterochromatin. Recruits a high density of cohesin. This
promotes bi-orientation of centromeres on the spindle, and prevents merotely
(because pombe has multiple microtubles per centromere, MT-binding sites must
be co-ordinated so they all face the same way - failure causes merotely and
lagging chromosomes).
In order to have any segregation function you must have a kinetochore. In order
have highly accurate segregation function you need the functions provided by the
centromeric heterochromatin. (This is in pombe, the role of pericentromeric
heterochromatin is less clear-cut in multicellular organisms.) In the absence of
centromeric heterochromatin there is premature separation of sister centromeres,
lagging chromosomes, etc. Segregation does not entirely fail because the
chromosome arms have cohesion (non-heterochromatin directed).
We have also shown that heterochromatin is required for establishment but not
maintenance of CENP-A chromatin.
The centromeres of Schizosaccharomyces are not conserved. No homology
found between e.g. pombe and octosporus. There may be some similiarity in
domain organisation - but that's in progress.
Anyway, hope that helps. I've included a variety of cartoons on pombe
centromeres which I hope helps in clarifying things.
Let me know if you have any further questions.
Cheers,
Alison
Many thanks if you can help
Best wishes,
Val