In the early 1980s, it was clear that specialized proteins must exist for accurate and regulated mRNA synthesis from particular genes in mammalian cells. However, nobody had been able to identify such “transcription factors” or determine how this process of transcriptional activation works. The breakthrough in my laboratory came when we found out that human cell extracts contained a factor that can discriminate between two templates and somehow program the enzyme that reads DNA to choose the right promoter DNA and ignore all others. But how?

We decided to use a short piece of the active promoter DNA sequence as “bait” to fish out proteins that selectively bind this site. The challenge was to purify this activity away from the other 3,000 DNA-binding proteins present in human cell extracts! After months of struggling with this problem, I vividly recall walking into the lab and my coworkers, Jim Kadonaga and Kathy Jones, saying, “We think we know which protein it is!” They had cleverly treated the human cell extract with a huge excess of sheared calf-thymus DNA to remove most nonspecific DNA-binding proteins, enriching the treated extract for the protein we wanted. Sequence-specific DNA affinity resin was then used to bind the transcription factor in the treated extracts, leading to the purification of a single protein. We called this protein specificity protein 1 (Sp1), the first of many sequence-specific transcription factors that were to prove critical for human gene regulation.

I’ll never forget the feeling of profound excitement at having shared the discovery of such a fundamental protein in biology and, at the same time, having devised with my lab members a new means to isolate hundreds more of these key gene-regulatory proteins.

—Robert Tjian, on discovering the first specific eukaryotic transcription factor

When complete eukaryotic genome sequences became available, molecular biologists were excited to discover the extent of the transcriptome, the entire set of RNA transcripts produced in a cell. Initially, researchers focused on characterizing the transcription products of known genes. These included mRNAs and known stable noncoding RNAs (ncRNAs) such as rRNAs, tRNAs, small nuclear RNAs (snRNAs) involved in pre-mRNA splicing, and small nucleolar RNAs (snoRNAs), which guide chemical modifications in the ribosome.

Unexpected levels of complexity began to emerge, however, beginning with the discovery of naturally occurring interfering RNAs, such as small interfering RNAs (siRNAs) and microRNAs (miRNAs), which have roles in the regulation of translation (as we describe in Chapter 22). Using a combination of microarrays and RNA-Seq (see Chapter 8), researchers could detect RNA transcripts without being biased by prior expectations. Use of these techniques showed that a lot of transcription was occurring that had previously been ignored. These new technologies revealed that the transcription landscape in higher eukaryotes is much more complex than expected. Surprisingly, a large fraction of transcripts originate from intergenic regions—regions between the coding sequences of genes—that had been thought to be silent, or from sequences that run in the opposite direction (antisense) to genes. Transcription that does not map to protein-coding genes or to known ncRNA genes also occurs in yeast.

In a parallel set of experiments, arrays of synthetic DNA oligonucleotides representing all nonrepetitive sequences in human chromosomes 21 and 22 were used to map the binding sites for three human transcription factors—Sp1, cMyc, and p53—that activate the transcription of many protein-coding genes involved in cell growth and differentiation. The experiments revealed far more transcription factor–binding sites than would be predicted from the number of protein-coding genes in these chromosomes. Of these binding sites, more than one-third lie within or immediately 3′ to well-characterized genes and seem to correlate with the transcription of ncRNAs. These findings have changed our thinking about transcription: much more of it goes on than previously suspected. Just what is all that RNA doing?

Some possible roles of previously undetected transcripts have emerged. For example, long noncoding RNAs (lncRNAs) are produced from regions that are either intergenic or antisense to genes. The functional significance of lncRNAs is not known, although several studies suggest roles for these transcripts in gene regulation. Shorter transcripts, particularly those that originate near gene promoters, fall into two somewhat arbitrary categories: molecules 20 to 200 nucleotides long are called small RNAs (sRNAs), and molecules of 200 to 1,000 nucleotides are called long RNAs (lRNAs). These categories include numerous subfamilies of transcripts defined by their abundance, longevity, and genomic origin. Although the sources of these transcripts are not yet fully worked out, at least some sRNAs may result from aborted or prematurely terminated transcription.

Whether pervasive transcription is simply a consequence of low-level background transcription or has functional significance is a subject of active research. In animals, it is unclear whether transcripts that are initiated near a particular promoter are functional. One possibility is that promoter-associated transcripts may help maintain chromatin in an open state that is more accessible to the transcription and regulatory machinery. In addition, it could keep a pool of RNA polymerase available for rapid deployment to make mRNAs. This will clearly be an expanding area of research, and more surprises are certainly in store.