University of Toronto press release distorts conclusions of RNA paper

My colleague, Ben Blencowe, just published a paper ...

Sharma, E., Sterne-Weiler, T., O’Hanlon, D., and Blencowe, B.J. (2016) Global Mapping of Human RNA-RNA Interactions. Molecular Cell, [doi: 10.1016/j.molcel.2016.04.030]

ABSTRACT (Summary)

The majority of the human genome is transcribed into non-coding (nc)RNAs that lack known biological functions or else are only partially characterized. Numerous characterized ncRNAs function via base pairing with target RNA sequences to direct their biological activities, which include critical roles in RNA processing, modification, turnover, and translation. To define roles for ncRNAs, we have developed a method enabling the global-scale mapping of RNA-RNA duplexes crosslinked in vivo, ‘‘LIGation of interacting RNA followed by high-throughput sequencing’’ (LIGR-seq). Applying this method in human cells reveals a remarkable landscape of RNA-RNA interactions involving all major classes of ncRNA and mRNA. LIGR-seq data reveal unexpected interactions between small nucleolar (sno) RNAs and mRNAs, including those involving the orphan C/D box snoRNA, SNORD83B, that control steady-state levels of its target mRNAs. LIGR-seq thus represents a powerful approach for illuminating the functions of the myriad of uncharacterized RNAs that act via base-pairing interactions.

The authors looked at RNA-RNA interactions (hybrid formation) by mapping different RNA molecules that bound to each other.

The idea is to map as many of these interactions as possible in order to get a handle on which of the RNAs have a significant biological function. The presence of an interaction does not prove that the RNA molecule(s) have a function since there are many cases where spurious transcripts could be complementary to an existing RNA. Only one of the RNAs might have a function or maybe neither has a function. However, it's one way of detecting possible functional RNAs.

The technique involves in vivo chemical crosslinking of the two interacting RNAs followed by sequencing. The authors tested the technique by showing that it detects known interactions between well-characterized RNA molecules. Then they applied it to all the interactions in human tissue culture cells.

A large fraction of the interactions involve associations between well-known functional RNA classes such as ribosomal RNAs (rRNAs), transfer RNAs (tRNAs), spliceosomal RNAs (snRNAs), snoRNAs, and messenger RNAs (mRNAs). These interactions don't make a significant contribution to our understanding of which genes in the genome are functional since even if every interaction proved there was a function it would still only amount to a very tiny amount of the genome.

The authors did detect a few interactions involving long noncoding RNAs (lincRNAs) but this was only a small percentage of the total. There are thousands of lincRNAs but currently there are only about 200 that have known functions and not all of these are even human.

It's the lincRNAs and other noncharacterized RNAs that represent the biggest problem because if all of them have a function then there will be >50,000 genes in the human genome instead of the current estimates of about 25,000 (20,000 protein-coding genes and 5,000 genes for functional RNAs). If all of those RNAs were functional they would occupy about 1% of the genome so this has has very little to do with whether 90% of our genome is junk.

This paper does not make any significant contribution to our understanding of pervasive transcription and whether most of the transcripts are functional (hint: they aren't). It does not make a contribution to the debate over junk DNA since most of the RNAs that were detected were already in the functional category or strongly suspected of being functional.

The authors make no claims about junk DNA in the paper because the paper is not about junk DNA.

Now let's look at the press release from the Donnelly Centre for Cellular and Biomolecular Research at the University of Toronto. This building is attached to the building that houses my department and the authors of the study are colleagues in a sister department (Department of Molecular Genetics). The press release was written by Jovana Drinjakovic, a science writer with a Ph.D. in neurobiology.

Shedding light on the ‘dark matter’ of the genome

What used to be dismissed by many as “junk DNA” is back with a vengeance as growing data points to the importance of non-coding RNAs (ncRNAs) – genome’s messages that do not code for proteins — in development and disease. But our progress in understanding these molecules has been slow because of the lack of technologies that allow the systematic mapping of their functions.

The actual paper shed almost no light on "dark matter" (whatever that is). The paper didn't mention junk DNA and it made no contribution whatsoever to the idea that 90% of our genome is junk. The paper was mostly concerned with interactions involving functional RNAs that have been known for decades. The press release grossly distorts the actual content of the paper in order to sensationalize the results and promote the Donnelly Centre.

This is unacceptable. It's time to put a stop to this nonsense. The actual paper was addressing the issue of how many of the noncoding RNAs are functional. It did not give us an answer. The press release makes it sound as though we are just learning about the existence of noncoding RNAs.

A plot of human RNA-RNA interactions detected by LIGR-Seq The new tool, called ‘LIGR-Seq’, captures interactions between different RNA molecules. When two RNA molecules have matching sequences – strings of letters copied from the DNA blueprint – they will stick together like Velcro. The paired RNA structures are then removed from cells and analyzed by state-of-the-art sequencing methods to precisely identify the RNAs that are stuck together.

“Most researchers in the life sciences agree that there’s an urgent need to understand what ncRNAs do. This technology will open the door to developing a new understanding of ncRNA function,” says Blencowe, who is also a professor in the Department of Molecular Genetics and Banbury Chair in Medical Research.

It's true that we would like to know how many of the RNAs are functional but it's not true that we are completely clueless. There's plenty of evidence supporting the idea that only 10% of our genome is functional. It's very likely that the vast majority of transcripts are spurious and nonfunctional. A good science reporter would put the study in proper context.

I send an email message to each one of the authors asking their opinion on how much of our genome is junk and whether they believe that their study contributes in any significant way to to the junk DNA debate. None of them replied.

I hope they're embarrassed about being misrepresented by Jovana Drinjakovic's press release. I suspect they're not embarrassed and that's a problem. Even if they actually believe that most of our DNA is functional, their peer-reviewed paper did not present any evidence one way or another. Press releases should not contain information and speculation that was not subjected to the peer review process that resulted in publication.

The press release was widely copied, e.g. Science Daily and EurekAlert (AAAS). The general public and other scientists are being give a totally false impression of the scientific consensus about junk DNA. Most knowledgeable scientists believe that the evidence supports the existence of large amounts of junk in our genome.

My colleagues Alex Pallazzo and his student Eliza Lee have addressed the issue of noncoding RNAs. They conclude that the default hypothesis has to be that most are nonfunctional unless there's solid evidence of function. Right now it looks like this evidence won't be found so most of these RNAs are spurious, they are junk RNA.

Palazzo, A.F. and Lee, E.S. (2015) Non-coding RNA: what is functional and what is junk? Frontiers in Genetics 6:1-11 [doi: 10.389/fgene/2015.00002]

The paper from the Blencowe lab didn't discuss these issues or reference the Palazzo and Lee paper.

---