70% of the genome is transcribed - then what?

This idea that most of your DNA is continuously transcribed has been floating around scientific circles.

I've blogged about it at least twice, and Coffee Mug at GeneExpression has mentioned it.

Keeping this in mind, here is some more interesting data from Danesh's lab.

Background: Certain small siRNA produced from the genes within the centromeric portion of S. pombe's chromosome will inactivate neighboring genes found in these regions. These siRNAs accomplish this task by binding to the RITS complex. In the new paper, Marc Buhler describes exactly how the RITS complex silences nearby genes and the answer is very surprising.

The bottom line: RITS uses the siRNAs to bind to newly made transcripts. Then RITS activates two processes that affect local genes:

1) It promotes the modification of nearby histones and thus contributes to "epigenetic"-type gene silencing of the centromeric genes
2) It recruits Cid14p and the TRAMP complex which is responsible for destroying abarent mRNA transcripts

The weird thing is that most of the gene silencing REQUIRES Cid14p or the second process. Without Cid14, there is no gene silencing. In other words the first process (i.e. histone modification) can't promote gene silencing without the second process! So gene silencing here is mostly an mRNA stability issue. In fact many papers have established a link between hoistone modifications and mRNA processing. So perhaps the state of the histones does not affect what RNA polymerase reads but rather how stable the RNA polymerase product is.

From the paper:

The high degree of conservation of both the heterochromatin machinery and RNA-processing pathways suggests that similar mechanisms may couple epigenetic gene regulation to RNA processing in multicellular eukaryotes.

So perhaps a large portion of the genome is constantly transcribed but histone modification dictates what portions of the genome yeild stable transcripts.

Ref:
Marc Bühler, Wilhelm Haas, Steven P. Gygi and Danesh Moazed
RNAi-Dependent and -Independent RNA Turnover Mechanisms Contribute to Heterochromatic Gene Silencing
Cell (07) 129:651-653 doi:10.1016/j.cell.2007.03.038

More like this

Well it would seem that in the past couple of years pop science has discovered RNA. Via Genomicron, I found this article in Scientific American from a few years back. Unfortunately all the lit on RNA in the popular press is a little overhyped and not very well understood. Sure, there is probably a…
Recently there has been a flood of press about epigenetics and non-coding RNA. What is lacking from these articles is a description of how DNA is packaged and what DNA elements such as promoters and enhancers do. Today I would like to touch upon all of these subjects with a post on how DNA is…
Here I am, in the lab with one last experiment to go before I leave to feast on a Christmas Eve dinner, so while I wait for that last centrifugation step, I'll write a quick post about all these great papers on RNA Polymerase II and chromatin remodelling. As I've said before, if you want to…
Well two weeks ago in Science, two reports came out about yet another species of small RNA ... rasiRNA ... uhm ... piRNA (OK they haven't harmonized their nomenclature yet). So here is a brief review of the types of RNA: - mRNA (messenger RNA). These are the RNAs that encode polypeptide chains. -…

I have no trouble believing that one can find bits of RNA complimentary to 70% of the genome if one screens enough cDNA libraries that have been enriched for very rare "transcripts."

That's not the real question. The real question is whether most of these "transcripts" are functional or whether they are rare spurious transcripts produced by accident.

Most of the predicted alternative splicing is artifact due to over-reliance on flawed EST data. You are in a good positon to comment on the validity of the other "transcripts." Are they real or are they artifacts as Struhl suggests?

OK here is more data that I haven't told you.

Last year the Brow lab published an RNA PolII Chip (Chromatin IP) with surprising results. Chromatin IP for those that don't know is a technique where you cross link all the proteins to the DNA that they are bound to in vivo. The protein of interest (here PolII) is isolated with an antibody directed against the protein (here PolII) and then you PCR amplify all the DNA that is co-eluted. So what is PolII bound to in yeast? Or rather how much of each genomic fragment is pulled down?

Let's say that the average active genes are pulled down with a certain frequency X. In comparison, the most active genes are pulled down with a frequency of only 4X. Regions of DNA with no active genes are also pulled down with a frequency of X. And silenced genes are pulled down with frequencies below background levels (>1X).

So it would seem that RNA Polymerase is sitting almost everywhere, but less so at inactive genes.

As for the Struhl paper - Kevin Struhl claims that the number of PolII molecules and the amount of DNA etc. in a typical yeast cell is consistent with this idea that much of the genome is transcribed at a low rate. The most important point of his article is not that all these transcripts are noise, but that this low level of transcription is happening.

Ref:
Eric J. Steinmetz, Christopher L. Warren, Jason N. Kuehner, Bahman Panbehi, Aseem Z. Ansari and David A. Brow
Genome-Wide Distribution of Yeast RNA Polymerase II and Its Control by Sen1 Helicase.
Mol Cell (06) 24:735-746

Struhl K.
Transcriptional noise and the fidelity of initiation by RNA polymerase II.
Nature Structural & Molecular Biology (07) 14:103-105

From the Steinmetz paper:

Perhaps the most striking outcome of the regression analysis [of the PoliII ChIP results] is that the x intercept, equal to the Pol II enrichment ratio at 0 transcripts per hour, is 1.0. This result indicates that the absence of Pol II enrichment corresponds precisely to the absence of productive transcription. Given that the enrichment ratio is well below 1.0 at regions that are silenced or repressed (Figure 3), this intriguing finding implies that genes that are producing no stable transcripts but are not actively silenced or repressed are associated with significant levels of Pol II.

With regards to are they "real" ... as in are the transcriptional products useful. Maybe some of them. But I bet that something else is going on. It looks like the actual act of transcribing the genome, including all the "junk DNA" is somehow important for some process.

Maybe the cell is analyzing all its genetic potential, maybe it has to do with heterochromatin structure, or perhaps the cell is scanning for DNA damage, which can be detected by PolII. But there is some piece of the puzzle missing.

"But there is some piece of the puzzle missing."

I think we're still looking for the corner pieces. I have something blue here, but can't tell whether it's sky or water.

Very cool story.

I think that the relevant issue is that chromatin modification may alter gene expression by regulating RNA stability. This idea is pretty radical and would be a major shift in how we view gene expression.

By Acme Scientist (not verified) on 23 May 2007 #permalink

AS,

I totally agree, this idea that RNA stability may play a crucial role in gene expression is very different then most current models. Also the idea that the inhibition of certain genes may involve a repression of this "basal transcription rate is also intriguing.

First, I forgot we we talking about yeast. There are about 7,000 genes of which 6,600 are likely transcribed by RNA polymerase II. The size of the genome is roughly 12,000 kb.

The average gene is larger than 1kb so it's not a big surprise that >70% of the yeast genome can be transcribed.

Second, all specific DNA binding proteins also bind non-specifically to any stretch of DNA. You can do the calculations if you know the parameters. I do this in my textbook for E. coli RNA polymerase and show that many (25%) of the molecules will, of necessity, be bound non-specifically at any point in time. The same calculations have been done for lac repressor and they show that for every molecule sitting on the lac operators there are ten that are bound non-specifically.

In E. coli, RNA polymearase probably finds a promoter by binding non-specifically to DNA then sliding along DNA until it encounters a promoter where it binds tightly. It can scan about 2000 bp before it falls off. (The process is called one-dimensional diffusion.)

It's likely that eukaryotic RNA polymerases work the same way so it shouldn't come as a big surprise that RNA polymerase II is sitting on lots of places in the genome. If the detection technique is sensitive enough then this would be the expected result.

I predict that if you do the same experiment with RNA polymerase III or RNA polymerase I you will get the same result. In fact, I predict you will get the same result if you express T7 RNA polymerase in yeast cells.

apalazzo says,

With regards to are they "real" ... as in are the transcriptional products useful. Maybe some of them. But I bet that something else is going on. It looks like the actual act of transcribing the genome, including all the "junk DNA" is somehow important for some process.

I think you're switching from yeast to mammals, right?

Just because a bit of DNA is copied into RNA does not mean that something important is going on. Based on what we know about the biochemistry of transcription and DNA binding proteins, it's quite reasonable to expect accidental random copying of junk DNA. There's so much of it that it's inevitable that RNA polymerases will be bound and it's inevitable that some of them will begin transcribing.

I'm not saying that this is the only possibility. What I'm saying is that it's the expected result, and therefore not surprising. We don't need to automatically assume that this is "somehow important for some process." It may be, but we need a lot more data before that becomes a reasonable hypothesis.

Maybe the cell is analyzing all its genetic potential, maybe it has to do with heterochromatin structure, or perhaps the cell is scanning for DNA damage, which can be detected by PolII. But there is some piece of the puzzle missing.

I don't see any pieces missing.

Larry,

I don't think that yeast and mammal chromatin structure will differ that much. But they do differ from E. Coli in that euks have histones and nucleosomes. The surprising finding of the ChIP experiment is that RNA Pol is just as likely to be siting on an active gene as it is on a random piece of DNA. What is surprising about the RITS paper is that parts of the genome that are "silenced" are actually transcribed. So what is the difference between gene and intergenic regions, or the difference between heterochromatin and euchromatin? It does not appear that DNA accecibility by RNA Pol II (the current model in the Euk transcription field) is the issue.