Kevin Struhl

Three fundamental classes of gene regulatory elements were identified in Escherichia coli in the 1960s: operators bound by repressor proteins, promoters bound by the basic transcription machinery, and enhancers bound by activator proteins. Promoters mediate constitutive gene expression, whereas operators and enhancers regulate expression in response to physiological conditions. As discovered in the 1980s, interactions between proteins associated with spatially separated elements can loop out the intervening DNA and regulate transcription. Eukaryotic gene regulation is mediated by the same three classes of genetic elements, but due to semantic confusion, it is often believed the eukaryotic enhancers activate transcription from long distances via enhancer–promoter loops. However, eukaryotic enhancers mediate only local changes in chromatin, and they stimulate transcription via directional and short-range interactions with the basic RNA polymerase II machinery at promoters. Enhancer action at a distance is mediated by loops between proximal and distal enhancers that bring activator proteins associated with distal enhancers in proximity to promoters. Thus, Jacques Monod's 1954 conjecture that “what is true for E. coli is true for the elephant, only more so” has proven correct.

The poly(A) profile of a typical yeast gene comprises ∼50 poly(A) sites corresponding to distinct 3′ mRNA isoforms. Poly(A) profiles are shifted upstream in strains with slow RNA polymerase II (Pol II) elongation rates resulting from Pol II mutations or from depletion of histone chaperones FACT or Spt6. Conversely, downstream-shifted poly(A) profiles occur in strains with fast Pol II elongation rates caused by Pol II mutations or by depletion of histones. Downstream-shifted poly(A) profiles also arise when components of the cleavage/polyadenylation machinery are depleted. Here we examine poly(A) profiles in 45 yeast strains with mutations that affect a wide variety of DNA- or RNA-based processes. Typically, altered poly(A) profiles occur in only a subset of genes, and some strains show both upstream and downstream poly(A) shifts depending on the gene. Hierarchical clustering of the poly(A) profiles reveals 6 classes of genes that behave similarly over the genetic conditions tested and have different RNA sequence preferences in their 3′ untranslated regions. Many chromatin-modifying or chromatin-associated proteins affect poly(A) profiles similarly to Pol II elongation rate mutants, indicating that they affect the rate of transcriptional elongation. Chromatin-modifying activities, but not cleavage/polyadenylation factors, often have non-uniform effects on 3′ mRNA isoform levels of the same gene, suggesting that chromatin differentially affects Pol II dwell time at individual sites during transcriptional elongation.

Although enhancers activate transcription from long distances, they stimulate transcription only through short-range interactions with the RNA polymerase II machinery. I posit that action at a distance is mediated by loops between distal and proximal enhancers that thereby bring proteins associated with distal enhancers near promoters. This Comment posits that enhancers stimulate the transcription machinery only through short-range interactions and, therefore, that enhancer activity at a distance is mediated by chromatin loops between distal and proximal enhancers, not between enhancers and promoters.

The Genetics Society of America's (GSA) Edward Novitski Prize is awarded to researchers for extraordinary creativity and intellectual ingenuity in genetics research. Struhl is being recognized for his pioneering work cloning a functional eukaryotic gene in E. coli, defining its promoter and regulatory region, and using random DNA and amino acid sequences to define determinants of specificity. The award also recognizes other key scientific contributions including Struhl's discovery of the sequences and protein interactions required for transcriptional activation and repression and demonstrating the importance of nucleosome-free regions for transcription initiation, among others.

Expression of a typical yeast gene results in ∼50 3’ mRNA isoforms that are distinguished by the locations of poly(A) sites within the 3’ untranslated regions (3’ UTRs). The location of poly(A) sites with respect to the translational termination codon varies considerably among genes, but whether this has any functional significance is poorly understood. Using hierarchical clustering of 3’ UTRs, we identify eight classes of S. cerevisiae genes based on their poly(A) site locations. Genes involved in related biological functions (GO categories) are uniquely over-represented in six of these classes. Similar analysis of S. pombe genes reveals three classes of 3’ UTRs, all of which show over-representation of functionally related genes. Remarkably, S. cerevisiae and S. pombe homologs share related patterns of poly(A) site locations. These observations suggest that the location of poly(A) sites within 3’ UTRs has biological significance.

The RNA polymerase II (Pol II) elongation rate influences poly(A) site selection, with slow and fast Pol II derivatives causing upstream and downstream shifts, respectively, in poly(A) site utilization. In yeast, depletion of either of the histone chaperones FACT or Spt6 causes an upstream shift of poly(A) site use that strongly resembles the poly(A) profiles of slow Pol II mutant strains. Like slow Pol II mutant strains, FACT- and Spt6-depleted cells exhibit Pol II processivity defects, indicating that both Spt6 and FACT stimulate the Pol II elongation rate. Poly(A) profiles of some genes show atypical downstream shifts; this subset of genes overlaps well for FACT- or Spt6-depleted strains but is different from the atypical genes in Pol II speed mutant strains. In contrast, depletion of histone H3 or H4 causes a downstream shift of poly(A) sites for most genes, indicating that nucleosomes inhibit the Pol II elongation rate in vivo. Thus, chromatin-based control of the Pol II elongation rate is a potential mechanism, distinct from direct effects on the cleavage/polyadenylation machinery, to regulate alternative polyadenylation in response to genetic or environmental changes.

Polyadenylation occurs at numerous sites within 3′‐untranslated regions (3′‐UTRs) but rarely within coding regions. How does Pol II travel through long coding regions without generating poly(A) sites, yet then permits promiscuous polyadenylation once it reaches the 3′‐UTR? The cleavage/polyadenylation (CpA) machinery preferentially associates with 3′‐UTRs, but it is unknown how its recruitment is restricted to 3′‐UTRs during Pol II elongation. Unlike coding regions, 3′‐UTRs have long AT‐rich stretches of DNA that may be important for restricting polyadenylation to 3′‐UTRs. Recognition of the 3′‐UTR could occur at the DNA (AT‐rich), RNA (AU‐rich), or RNA:DNA hybrid (rU:dA‐ and/or rA:dT‐rich) level. Based on the nucleic acid critical for 3′‐UTR recognition, there are three classes of models, not mutually exclusive, for how the CpA machinery is selectively recruited to 3′‐UTRs, thereby restricting where polyadenylation occurs: (1) RNA‐based models suggest that the CpA complex directly (or indirectly through one or more intermediary proteins) binds long AU‐rich stretches that are exposed after Pol II passes through these regions. (2) DNA‐based models suggest that the AT‐rich sequence affects nucleosome depletion or the elongating Pol II machinery, resulting in dissociation of some elongation factors and subsequent recruitment of the CpA machinery. (3) RNA:DNA hybrid models suggest that preferential destabilization of the Pol II elongation complex at rU:dA‐ and/or rA:dT‐rich duplexes bridging the nucleotide addition and RNA exit sites permits preferential association of the CpA machinery with 3′‐UTRs. Experiments to provide evidence for one or more of these models are suggested.