Question: "Coding" Epigenetic Elements
by JoySD published an article entitled Structure of Key Epigenetics Component Identified that hails the reporting of the 3D structure of a key protein, UHRF 1, in ensuring the "epigenetic code" (attached chemicals for methylation) is accurately copied when the DNA is replicated.
A question for those here who claim to understand all these details (and have access to Nature, please. How do chemical attachments to DNA get copied in this process? Does the protein somehow get reverse-engineered into standard DNA code, or does the replicated string come with non-nucleotide chemical 'markers'? Does the DNA now encode this particular protein in this particular spot (standard DNA code)?
Some clarification of what, exactly, is being copied, "proof-read" and reconstructed in the daughter cell would be appreciated. Thanks to anyone who can translate this for me.
September 9th, 2008 at 1:18 pm
Joy:
Nope^3. DNA gets methylated (a CH3 being attached to a CpG dinocleotide). The methylation pattern is the "epigenetic code", so to speak. During DNA replication, the methylation pattern gets copied by a particular DNA methyl transferase (DNMT). Another enzyme (UHRF1), the one whose structure was published this week, targets the DNMT to the replication fork (which is hemi-methylated at this point because one of the strands doesn't have the methyl group yet). As far as I understand it. No magic involved.
Comment by Raevmo — September 9, 2008 @ 1:18 pm
September 9th, 2008 at 2:14 pm
Raevmo:
I know. I'm asking *what* about this gets "encoded," and *what* it means to be "coded" in the DNA template.
So… is it little "insert methyl group here" notations (like holds as dots on sheet music notes) that gets "encoded?" Are these little notations "encoded" in standard DNA language or something else? Is it a different kind of coding?
Stop inserting stupid little barbs that serve only to make your posts objectionable. Try to stick to the topic.
I am asking a strictly mechanical question. In preparation for cell division, the DNA gets copied. The double helix with its coded nucleotides is untwisted and each strand serves as a template for producing its new 'other strand' to produce two sets of chromosomes.
I am asking if the methylation information ("epigenetic code") is translated into "genetic code" (nucleotide sequences translating to "insert here" or whatever) or if the protein itself is synthesized as part of the replication process and attached to backbone of the 'new' DNA helixes in its original spot.
IOW, is the process of replicating genomes concurrently attaching pre-synthesized proteins to the DNA? I'm trying to figure out what - exacly - is written in actual "Code" here. We know DNA is written in code. Are the expression-related non-DNA attachments to DNA also written in code, are they synthesized as a subprocess of normal DNA replication (therefore NOT "encoded"), or are they just grabbed from the nuclear environment and attached to the new strands where they were attached to the old strands (therefore NOT "encoded" or synthesized)?
It may seem like a dumb question to you, but not to me. The press release looks to be written for people who already know how this is done, and I am not among those people. So I'm trying to get a fair layman's description of this process.
Comment by Joy — September 9, 2008 @ 2:14 pm
September 9th, 2008 at 2:54 pm
> am asking if the methylation information ("epigenetic code") is translated
> into "genetic code" (nucleotide sequences translating to "insert here" or
> whatever) or if the protein itself is synthesized as part of the replication
> process and attached to backbone of the 'new' DNA helixes in its original spot.
Apologies if this spells out things you already know. I wanted to make sure everything is clear.
No, the methylation isn't not translated into "genetic code." As you say, when DNA is replicated, the old strands are separated and used as template to synthesize a new strand.
The epigenetic information is the methyl groups attached to cytosine residues. Typically, methylation "silences" DNA by preventing transcription, so if you are thinking in terms of information or code, methylation means "don't express this." Since different cells express different genes, the pattern of methylation will be different in different cell types and it is important to maintain that pattern when cells replicate.
When the DNA is replicated, the new strand is built using unmethylated cytosines. The DNA is now hemi-methylated — methylated on only one strand. In order for the epigenetic information to be successfully passed on to the daughter cells, the new strand of DNA must be methylated in the same way as the parent strand . Not all CpG dinucleotides are methylated, so a process is needed to selectively methylate the same nucleotides in the daughter strand as are methylated in the template.
At the time of DNA replication, UHRF1 is already floating around in the nucleus. The details of UHRF1 structure are above my pay grade, but from what I can understand, the UHRF1 protein contains a pocket whose size and other biochemical features allow it to bind methylated cystosines in hemi-methylated DNA. UHRF1 does not bind to unmethylated cytosines. The UHRF1 also binds to a DNA methylase. Since the DNA methylase is now in close proximity to the unmethylated cystosine of the daughter strand, it can now transfer a methyl group to the unmethylated base in the daughter strand. For bases that are unmethylated in the parent strand, UHRF1 will not bind, and the equivalent base will not be methylated in the daughter strand.
Comment by Nick — September 9, 2008 @ 2:54 pm
September 9th, 2008 at 3:23 pm
One quick clarification, suggested by your last comment in the EAM thread. Methyl groups are not protein.
Comment by Nick — September 9, 2008 @ 3:23 pm
September 9th, 2008 at 3:30 pm
Joy,
It seems to me Nick explained the mechanics very well (but what do I know).
What protein do you mean? A methyl group is not a protein.
The interesting question is, why do some CpG's get methylated de novo and others don't? How is that encoded? Perhaps that is what you meant?
I find this interesting because one of the things I study (well, not me really, one of the PhD students that I advise) is the sexual development of certain scale insects. Many species of scale insects don't have X and Y chromosomes, but instead it appears that the mother facultatively somehow causes imprinting (by methylation) of the chromosomes of some of her eggs, turning them into sons. The rest becoming daughters. The funny thing is that during spermatogenesis those sons destroy the DNA they get from their father, thus only passing on their maternal DNA to their sperm. Seems like mom has a very clever strategy to pass on her genes at the expense of her partner's genes.
Comment by Raevmo — September 9, 2008 @ 3:30 pm
September 9th, 2008 at 3:44 pm
Thank you, Nick. That's just what I was looking for.
Comment by Joy — September 9, 2008 @ 3:44 pm
September 9th, 2008 at 3:54 pm
Raevmo:
Right, it's a chemical attachment. But UHRF1 *is* a protein, apparently has a selective affinity with one-sided attachments and completes methylation on the 'new' side when the new double-strand is done (if I understand what Nick said). It apparently is NOT synthesized along with the other strand during replication, but is recruited from the nuclear environment (?) to "mark the place" where a methyl group should be and transfer another group to the replicant strand, opposite.
How much UHRF1 is usually present in the nuclear environment, and what's it's job when replication isn't happening?
Comment by Joy — September 9, 2008 @ 3:54 pm
September 9th, 2008 at 4:20 pm
Joy:
UHRF1 helps another protein (the DNMT) to methylate the unmethylated (both strands are new in a sense, but only one is unmethylated). Or something like that.
Proteins are not synthesized along with DNA. I'm surprised you didn't know that. How can you have such a strong opinion about EAM when you don't know basic stuff like this?
No idea how much is present (why does it matter?) It probably doesn't have another job (but who knows).
Comment by Raevmo — September 9, 2008 @ 4:20 pm
September 9th, 2008 at 4:39 pm
Apparently, UHRF1 levels are low in quiescent cells, and it is dramatically upregulated in replicating cells. That would make sense for a protein whose primary function is in DNA replication.
Interesting tidbit: Since cancer cells replicate like crazy, they tend to have high levels of UHRF1. There are several papers suggesting that UHRF1 could be a good target for anti-cancer drugs. That seems a little counter-intuitive to me. If cancer is associated with aberrant DNA methylation, then I'd think the last thing you'd want to do is down-regulate a gene that is required for proper methylation. I guess the idea is that if you can inhibit enough of the genes involved in DNA replication, tumor cell division will grind to a halt.
Comment by Nick — September 9, 2008 @ 4:39 pm
September 9th, 2008 at 5:21 pm
Hi Nick,
If I understand it correctly, UHRF1 facilitates methylation during replication. But cancer cells are associated with a hypomethylated state especially around oncogenic regions (+- 300 oncogenes at present e.g. k-ras EGFR). So targeting the UHRF1 protein for inhibition might be a bad idea, but targeting it to increase its activity (via directly or indirectly) might be a better idea, resulting in proper methylation and possibly proper "reprogramming" of cancer cells during replication. So basically the idea is not to kill cancer cells by selectively inducing programmed cell death, but reprogramming cancer cells to start acting like non-cancerous cells again?
Also, do you know anything about the mechanism and process during meiosis that causes the cells to become totipotential cells? Does epigenetic reprogramming play a role, and are epigenetic changes picked during life also imprinted during this process?
Thanks in advance
Comment by Telicmeme — September 9, 2008 @ 5:21 pm
September 9th, 2008 at 6:48 pm
Raevmo:
Wasn't I just informed that proteins aren't involved in methylation of genes? I think I can picture the UHRF1 handy-binding the one methylated strand while capturing and releasing a stray methyl group to 'transfer' to the other strand. Now it's two proteins? "Something like that" sounds about right…
A ribosome on the nuclear envelope could be triggered to kick in when the first replication alarms go off, basically producing UHRF1 (and whatever other particular proteins serve the epigenetic function during replication). That doesn't seem odd at all. Which is why I asked, if it's not produced at the same time the replicant strands are produced, what its 'other' function(s) might be. Lots of well-oiled machinery kicks in during the process of replication, I was asking about how well-oiled the epigenetic inheritance machinery is, in this picture.
If you can't refrain from rudeness, you'll find yourself out of this thread, too.
Comment by Joy — September 9, 2008 @ 6:48 pm
September 9th, 2008 at 7:01 pm
Nick:
Thanks again, Nick! Yes, that does make sense.
Comment by Joy — September 9, 2008 @ 7:01 pm
September 10th, 2008 at 9:29 am
Hi Telicmeme,
I think the problem with that idea is that the UHRF1-DNMT1 complex doesn't know what the methylation state should be. It simply replicates the existing methylation state of a parent DNA strand onto the daughter strand. So, if a cancer cell is hypomethylated, overexpressing UHRF1 will ensure that the daughter cells accurately reflect the methylation state of the cancerous parent, efficiently making more cancer cells. In other words, if a cancer cell has lost methylation at oncogene X, then when it replicates, UHRF1 will not re-methylate the unmethylated sites in X.
Comment by Nick — September 10, 2008 @ 9:29 am
September 10th, 2008 at 11:37 am
Hi Nick,
Thanks for that. I'm still getting into this whole "methylcytosine = 5th base" epigenetics story, please forgive my ignorance. I understand now where my misunderstanding was ( I hope). Even though cancer cells are characterized by hypomethylation, they are also characterized by promoter-specific hypermethylation of cancer-associated genes, especially tumor-suppressors (e.g. p53). Hypermethylation of tumor suppressors would thus silence gene expression of these genes and promote oncogenic transformation.
So inhibiting the activity of the UHRF1-DNMT1 complex in theory might decrease hypermethylated tumor-suppressor genes, and hopefully reactivate them. In addition, increasing the already hypomethylated cancer epigenome might cause further genetic instability, and hopefully not predispose it to further transformation. Getting there?
I am interested in the mechanism and process during meiosis that causes the cells to become totipotential cells (currently doing the pubmed thing)? I am also interested in the mechanism through which differentiated (e.g. fibroblasts) cells can be reprogrammed to become pluripotential cells.
Surely if a mechanism is discovered to reprogram differentiated cells (e.g. fibroblasts), such a mechanism can be applied to cancer cells in order to reprogram them to their normal state?
So much to read, learn and discover…
Comment by Telicmeme — September 10, 2008 @ 11:37 am
September 10th, 2008 at 2:00 pm
Nick (or anyone)
Is there a mechanism for going in other directions? I.e., so that a methylated guanine on one strand will cause the corresponding cytosine to get methylated? If not, you are left with a replicating strand with methylated cytosines, and a non-replicating (i.e., replicates without the epigenic information) strand.
Is there a de-methylating process, such that either guanine or cytosine lose a methyl group (analogous to the original methylation)? Is there then a mechanism for that to be replicated in both directions? If not, has DNA become progressively more methylated over the last 4 billion years?
Comment by The Pixie Again — September 10, 2008 @ 2:00 pm
September 10th, 2008 at 4:13 pm
Pixie,
Ahh, I guess that was one bit I didn't make clear in my original summary. There are no methylated guanines. Only cytosine is methylated, and it is only methylated in the context of a CG dinucleotide (or CpG where p=the phosphate connecting the two bases). Since the reverse complement of CpG is CpG, the target for methylation is present on both strands. However, the actual methylated base will be shifted by one position relative to the one on the other strand — the methylated C will be opposite the unmethylated G on the other strand.
There is a de-methylation process, but I don't know much about it. Methylation is dynamic; different cell types with different methylation patterns are all derived from a single cell embryo. During meiosis, methylation is stripped from the DNA and then re-applied. This is particularly important for imprinted genes which are methylated differently depending on whether they is inherited from mother or father.
As for whether the genome has become progressively more or less methylated, there's an interesting twist. Methylated cytosine can undergo a process called spontaneous deamination which converts the C to a T. This mutation tends to be missed by the DNA repair machinery, so a C that is usually methylated is likely to be lost during evolution, unless there is selective pressure to maintain it. Indeed, the human genome contains fewer CpG dinucleotides than you would expect if the nucleotides were randomly distributed. The CpGs that are present tend to be clustered in so-called CpG islands. CpG islands are frequently found in the promoters of genes and are often unmethylated or methylated only in tissues that do not express the gene. Presumably the CpGs in promoter CpG islands remain unmutated because a) they are often unmethylated or b) where methylation is playing a dynamic role in gene regulation, there is selection against mutations that would destroy the methylation site.
Comment by Nick — September 10, 2008 @ 4:13 pm
September 10th, 2008 at 4:20 pm
One more clarification. Methylation is limited to CpG in mammals. I don't know about other animals or plants. Bacteria methylate C in different contexts and can also methylate A.
Comment by Nick — September 10, 2008 @ 4:20 pm
September 10th, 2008 at 4:37 pm
Nick
Thanks, that makes sense now.
Comment by The Pixie Again — September 10, 2008 @ 4:37 pm
September 10th, 2008 at 4:53 pm
As Nick points out, methylization mostly concerns developmental specialization and is reset during gametogenesis. How the gamete is imprinted is still rather mysterious, but may be sensitive to the environment.
Of note, there is also a sexual genetic conflict between the paternal and maternal lines. In placental mammals, if the male can direct the embryo to acquire additional nutrients from the mother, then it can be advantageous to the male line. In response, the paternal imprinting may be partially erased soon after fertilization by the oocyte. In other words, it is apparently consistent with selection.
Comment by Zachriel — September 10, 2008 @ 4:53 pm
September 10th, 2008 at 6:59 pm
Zach:
Specific imprinting of genes in gametes is one of those "War of the Sexes" things. The ability to alter that imprinting is "apparently consistent" with the kind of selection that works by design, but not the kind of selection that results in 2/3 of all pregnancies ending by spontaneous abortion.
The ability (and means) to alter the heritable epigenetic coding in gametes and/or zygotes looks a lot more like EAM than NDS. Are the changes "random" in cause, sequence and/or effect?
Comment by Joy — September 10, 2008 @ 6:59 pm
September 10th, 2008 at 9:53 pm
In placental mammals, spontaneous abortion is a every effective means of reducing the strain on the mother, such as when carrying a fetus with a significant developmental defect or when the mother is under environmental stress. This is the opposite case of male imprinting, which can increase the size of the fetus, enhancing its fitness at some expense to the mother. This is consistent with evolutionary selection due to genetic conflict between the sexes.
EVOLUTION OF SEX: A Genomic Battle of the Sexes, Science 1998
Evolution of imprinting mechanisms: The battle of the sexes begins in the zygote, Nature 2001.
Comment by Zachriel — September 10, 2008 @ 9:53 pm