According to the textbook model of gene transcription, the process is sparked by the recruitment of RNA polymerase II (Pol II) to a gene’s promoter. But a 2007 Cell paper from Richard Young’s lab at the Whitehead Institute in Cambridge, Mass., suggests the picture may be significantly more complicated at many genes, particularly those encoding key regulators of development.

That study, a Hot Paper this month, took a genome-wide look at transcription initiation in human embryonic stem cells. The authors found evidence for transcription initiation—that is, Pol II was recruited and the right chromatin marks were present—in the majority of all protein-coding genes. This was a perplexing result, since no more than 50% of such genes were thought to be undergoing transcription.¹ In other words, according to the paper, transcription is initiated in most protein-coding genes, but only a subset of those are actually transcribed. Some additional mechanism, then, must cause Pol II to stall after recruitment at some genes, yet continue transcription at others.

This proposed “a fundamentally different view” of how transcription of some genes is regulated—specifically, it means that “the binding of polymerase is not the rate-limiting step,” says Mike Levine of the University of California, Berkeley. “That was a very surprising and somewhat rude awakening.” Indeed, says Young, the finding “was such a surprise that we [checked] whether it was some strange element of the methodology or the cells we were studying.”

The study, and the genome-wide approach it used, is fueling a kind of revival in the field, says Julia Zeitlinger at the Stowers Institute for Medical Research in Kansas City, Mo. First, if initiation isn’t the key step in transcription at these genes, what is? One possibility is the transition from initiation to the elongation of the transcribed RNA, a process that’s now being more intensely studied. Researchers have also begun to scrutinize what stalled Pol II is doing, and proposed a plethora of hypotheses for why transcription might pause after recruiting the necessary machinery, which some evidence has already started to support.

Just a few months after the Cell paper was published, Young’s and Levine’s labs published a study in Nature Genetics that nailed down where in initiation the process gets held up: Pol II, they found, was sitting atop the promoter, stalled but ready to go.² “It’s like the engine is on, but the transmission is in neutral—just waiting to shift into gear,” says Levine. Stalling, too, occurred with especially high frequency on developmental genes, suggesting that stalling plays a role in the complex regulation of development, says Zeitlinger, the paper’s first author. “Understanding the control of this pause may be tantamount to understanding a key part of development,” says Young.

Stalling was not a new phenomenon: John Lis of Cornell University in Ithaca, NY, demonstrated Pol II stalling in genes coding for heat shock proteins 2 decades ago. The mechanism made sense for heat shock genes, which must be poised to switch on quickly in response to environmental or toxic stress. Although he had some evidence that the mechanism may be more general,³ says Lis, many in the field considered it a “peculiarity” specific to a small group of genes.

The wider role of stalling, though, “is still an open question,” says Matthew Lorincz at the University of British Columbia. It may, as in heat shock genes, be prepping the system for quick gene activation by keeping the chromatin in an active state. Alternatively, Pol II may be acting as a flag that helps transcription factors more easily locate genes, or as a bookmark to keep genes available for later use. “My take on it is that it serves to protect these promoter regions against silencing activity” by, for example, preventing DNA methylation at those sites, Lorincz says.

In a recent paper in Science, Levine’s group suggests yet another role for stalling: stalled polymerases are activated in a synchronous manner, the study shows, whereas the genes that lack the stalled polymerase come on in a stochastic fashion.⁴ “Having synchrony means having precision”—crucial to ensure that developmental processes occur in the order they’re supposed to, Levine says.

Another interesting point of the Cell paper, notes Lorincz, was that the stalled Pol II molecules do manage to produce very short transcripts which may play a regulatory role as well. “Every time people have looked carefully at small RNAs they’ve ended up having a regulatory role,” notes Young, so “it wouldn’t be a surprise” if that were true here too.

Last year, back-to-back Science papers—one from Lis’s lab and the other a collaboration between Young and Phillip Sharp at the Massachusetts Institute of Technology—added another twist to transcription, demonstrating that the polymerase, when it does shift into gear, can move bidirectionally, in either the sense or the antisense direction.^5,6 Normally, as Pol II travels down the strand, the DNA strands separate in front of it, exposing naked DNA. Two Pol II molecules starting back to back and traveling for short distances in opposing directions exposes even more naked DNA, which may benefit the cell by rendering the DNA more open to regulation, Young speculates.

“A few years ago, we thought that transcription was solved,” says Zeitlinger, but over the past couple of years, “it became clear that there are a lot of things we don’t understand.”

Data derived from the Science Watch/Hot Papers database and the Web of Science (Thomson ISI) show that Hot Papers are cited 50 to 100 times more often than the average paper of the same type and age. M.G. Guenther et al., “A chromatin landmark and transcription initiation at most promoters in human cells,” Cell, 130:77-88, 2007. (Cited in 215 papers)

As other comments indicated, the existence of abortive initiation complexes was well-known, certainly to those in the field. What the genome-wide studies did was to show the prevelence of abortive initiation in a non-biased survey. So, depending on your pre-conceived notions, the broad use of this mechanism was perceived as either not suprising or suprising.

We scientists routinely build theories based on the evidence at hand. If the evidence is insufficient to make general conclusions, most of us incorporate that as a tentative model but remain aware that the model hinges on a shaky pillar. So, one often designs studies with some element of intuition, based on partial evidence. That part of the fun of conducting science, which inherently deals with unknown variables. This is resolved when one of our colleagues, or ourselves, takes it upon her/himself to test that notion. Our best guesses become substantiated or refuted and we modify our studies accordingly.

Scientific progress is a wonderful process that I personally feel lucky to participate in. Am I sometimes suprised? Yes!! But I always know that there are others who, faced with the same poor evidence, arrived previously at what ultimately was the correct conclusion.

There are lots of things to learn and appreciate in this new wave of publications that address polymerase stalling. However, it is really not at all a surprise to those who are in this field. I am glad that at least John Lis' earlier work got a mention. But there were other reports more than 15 years ago that reported that polymerase pausing is not just a specialty of a few of genes but that it is a general rate-limiting step of transcription that occurs independent of transcriptional activators and transcribed sequences (for example PMID:7698646).

A general role for post-initiation control has been known for awhile.

Not just HSPs two decades ago. Marcu, et al., were reporting what they called "attenuation" in the myc promoter in the 80s. They had a theory that DNA secondary structure in exon 1 played a role.

Hi Susan -- thanks for reading. Hot papers are almost always a couple of years old, just because it generally takes that long to accumulate citations. -Alla Katsnelson

I would like to to know if this phenomenon was observed in single stranded DNA viruses. "Transcription Surprise" could be a mechanism whereby the cell chooses either the maternal strand or paternal strand of the double helix. Perhaps the partial copy stops the process from wasting or making both maternal and paternal RNAs.

"A few years ago we thought transcription was solved."

Oh really? Only someone who is not in this field would say such a thing. We know a great deal about how transcription works in vitro, and on naked DNA templates using TATA box promoters. But we don't know the first thing about how transcription works on non-TATA promotors (the majority of promoters in the genome), nor about how transcription proceeds through genes still coiled in chromatin, nor how it works in vivo. That's an awful lot that hasn't been solved, n'est-ce pas?

Interesting, isn't it, that the "hot paper of the month" was published TWO YEARS AGO. I suppose it's a good thing that someone has noticed this work (better late than never), but for those of us toiling in the transcriptional trenches, this is no surprise at all.