Transcription requires the pre-initiation complex to be assembled at the promoter
Humans have four RNA polymerases that make RNA copies of a DNA template. Pol-mt transcribes only mitochondrial DNA; Pol I is specialized for the sequences encoding the main ribosomal RNAs; and Pol III transcribes the genes for some of the numerous small noncoding RNAs. In this chapter we are concerned with RNA polymerase II, which transcribes all protein-coding genes and most noncoding RNA genes. As described in Chapter 1, transcription starts with assembly of a pre-initiation complex. First TFIID and TFIIA bind to the core promoter, followed by TFIIB, TFIIF, and Pol II, and finally TFIIE and TFIIH. Figure 1 showed the sequence motifs of a canonical core promoter—TATA box, Inr, DPE, and BRE—to which one could add the CCAAT box at −50 to −80 nt relative to the transcription start site. There is a nucleosome-free region of around 150 bp, and flanking nucleosomes carry H3K4me3 near active promoters, or H3K27me3 at repressed promoters. In undifferentiated cells a proportion of promoters are enriched for both the active H3K4me3 and the repressive H3K27me3 mark. It is thought that these bivalent or poised promoters are inactive but poised to become active as the cell differentiates. ENCODE identified 70,292 sequences in the human genome with promoter-like chromatin signatures.
Fig1. Consensus sequences for some core promoter elements often found in genes transcribed by RNA polymerase II. The TATA box is bound by the TATA-binding protein (TBP) subunit of transcription factor TFIID. The initiator (Inr) element defines the transcription start site (the A highlighted in red) when located 25–30 base pairs (bp) from a TATA box. The downstream core promoter element (DPE) is only functional when placed precisely at +28 to +32 bp relative to the highlighted A of an Inr element. Both Inr and DPE are bound by TFIID. Transcription factor TFIIB binds to BRE (TFIIB recognition element) and accurately positions RNA polymerase at the transcription start site. However, none of these elements is either necessary or sufficient for promoter activity, and many active RNA polymerase II promoters lack all of them. N represents any nucleotide. (Modified with permission from Smale ST & Kadonaga JT [2003] Annu Rev Biochem 72:449–479; PMID 12651739. © 2003 by Annual Reviews, http://www. annualreviews.org.)
However, few real promoters completely fit this standard description. As mentioned above, about 70% of promoters are associated with CpG islands, and these tend to lack most of the core elements. Promoters differ in the precision with which they specify the transcription start site. Focused or sharp promoters have one or a few fixed transcription start sites. They tend to have TATA, Inr, BRE, and DPE motifs, and they tend to regulate transcription of tissue-specific or signal-responsive genes. Dispersed or broad promoters allow transcription to initiate at many positions over a stretch of maybe 100 nucleotides. They are less likely to show any of the canonical core elements apart from multiple Inr related motifs. The genes they regulate tend to be housekeeping genes that are expressed at a relatively steady level in many cell types (Figure 2).
Fig2. Canonical and noncanonical core promoters. (A) A canonical promoter showing the TATA box, upstream (u) and downstream (d) BRE elements, and Inr, TCT, DCE, and DPE elements. TCT is used by the TATA-binding protein-related factor TRF2 in ribosomal protein-coding genes. See Figure 1.16 and Roy & Singer (2015) for descriptions of these elements. (B) A noncanonical promoter marked by CpG islands and a region depleted in ATG sequences. TSS, transcription start site; NFR, nucleosome-free region. (Adapted from Roy AL & Singer DS [2015] Trends Biochem Sci 40:165–171; PMID 25680757. With permission from Elsevier.)
Contrary to the simple picture, it would appear that many promoters, especially those lacking TATA boxes, are bidirectional, initiating divergent transcription from both DNA strands in opposite directions. Only a small proportion of the transcripts are stable polyadenylated mRNAs; many are part of the pervasive transcription identified by ENCODE. Also, many promoters are occupied by RNA polymerase II in a pre-initiation complex even when the downstream gene is not being expressed. It appears that the decisive control on gene expression is not so much assembly of the pre-initiation complex, as whether the polymerase is able to move into elongation mode, as described below. Roy & Singer (2015) (PMID 25680757) review the whole topic.
Many genes have more than one promoter
At least half of all mammalian genes have two or more alternative promoters. These will drive transcription from alternative versions of a first exon, which may or may not be coding. introns are defined by consensus sequences at either end: a 5′ donor sequence including the invariant GT (GU in the RNA), and a 3′ acceptor sequence with the invariant AG. Alternatively viewed, exons are flanked by 5′ acceptor and 3′ donor sequences. However, the 5′ end of the first exon does not have an acceptor splice site. Thus splicing of the transcript from an upstream exon 1 can pass over the sequence of any number of downstream alternative first exons to splice on to exon 2 (Figure 3). Some genes have whole batteries of alternative first exons. The UDP glycosyltransferase gene UGT1A1 on chromosome 2 (position 2q37) has 13 alternative first exons. On chromosome 5, the protocadherin α and γ genes each consist of large tandem arrays of 2400 bp alternative first exons that encode the bulk of the protein. These are spliced onto three small invariant exons that encode the C-terminal part of the protein. There are some analogies to the immunoglobulins, in that both have variable N-terminal but constant C-terminal regions, although this is achieved by entirely different means.
Fig3. A gene with several alternative promoters. This gene has four alternative promoters (Pα–Pδ ) and first exons (1α–1δ , shown as different-colored boxes), plus two downstream exons (E2, E3; pink boxes). The positions of splice donor (D) and splice acceptor (A) sites are shown. Each splice junction has to be made by joining a donor and an acceptor site, so exon 1α is spliced on to exon E2 and not exon 1β . The mature mRNA has a poly(A) tail.
Some alternative promoters are internal to a gene, within an intron, rather than upstream. In such cases, exons upstream of the internal promoter are not included in that transcript. The 79-exon dystrophin gene has several examples (Figure 4).
Fig4. Alternative promoters for the dystrophin gene. The positions of seven alternative promoters are shown at the top. C, cortical; M, muscle; P, Purkinje cells; R, retinal, brain, and cardiac muscle cells; CNS, central nervous system; S, Schwann cells; G, general (almost ubiquitously expressed, but not detectable in fully differentiated skeletal muscle). Each promoter uses its own first exon, shown in red on the diagram of the gene structure, together with the shared downstream exons shown and numbered in blue. The protein product from each is named for its molecular mass; for example, Dp427 is a 427 kDa dystrophin protein.
The alternative promoters can serve two purposes. First, they can involve different regulatory elements. Tissue-specific alternative promoters can allow different regulation of gene expression in different tissues. The various promoters in the dystrophin gene are an example. Differential regulation may be specific to particular developmental stages. Even imprinting can be differentially regulated; for example, the PLAG1 gene on chromo some 6 has two promoters located 55 kb apart. The downstream promoter is imprinted and is responsible for imprinted expression of the gene in many tissues. However, the upstream promoter is not imprinted, and allows biallelic expression in peripheral blood leukocytes. It is perhaps significant that the first intron of a gene is often by far the longest (genome-wide mean size 14,186 bp, compared to 4,847 bp for internal introns). Thus, promoters often lie far upstream of the internal exons of a gene, and alternative promoters are often well separated from each other. This would allow promoters to sit in a different chromatin environment from each other and from the bulk of the gene.
A second value of alternative promoters is that they allow functionally significant sequence differences between alternative first exons. Alternative 5′ untranslated regions may contain different regulatory elements. If the alternative exons contain coding sequence, they can result in protein isoforms with different properties. The examples of UDP glycosyltransferase and protocadherin, described above, show how this mechanism can allow a single gene to generate a whole family of proteins. Isoforms may have different subcellular locations (for example, soluble or membrane-bound) or different functions. For example, the progesterone receptor gene PGR on chromo some 11 uses alternative promoters to produce two isoforms, PRA and PRB, that differ by 165 N-terminal amino acids. Both isoforms are transcription factors, members of the nuclear receptor family described in Chapter 3, but they target different response genes and have different physiological effects. An extreme example is the CDKN2A gene. Here, two alternative first exons each contain a translational start site. Depending on which one is used, the sequence of the shared downstream exons is translated in different reading frames. Thus, the same downstream exons encode totally different proteins, depending on which promoter is used.
A quick check with any genome browser shows that most human genes encode more than one transcript. In December 2017, ENSEMBL listed 200,310 human transcripts but only 20,338 protein-coding genes. The ENCODE pilot project identified 2608 transcripts from the 487 loci examined, an average of 5.4 per locus. In addition to the use of multiple promoters, most transcripts are subject to alternative splicing, as described below. An additional mechanism, RNA editing, which operates only on certain genes, is also mentioned below. A further example, the remarkable series of complex rearrangements that generate the vast diversity of immunoglobulin and T-cell receptor molecules from a small number of genes.
Enhancers
Enhancers are regulatory elements that are located some distance away from the gene whose expression they control. They may be upstream or downstream of their target gene and may be as far as a megabase away. In some cases there may be other genes located in the DNA between an enhancer and its target. Enhancers have much in common with promoters—both are regions of accessible chromatin containing binding sites for transcription factors. Enhancers tend to carry the H3K4me1 mark on associated nucleosomes, compared to H3K4me3 on promoters. Sequences that carry the chroma tin signature of enhancers in one cell type may lack this in other cell types, and may not work in them in functional studies. Overall, ENCODE identified 399,124 enhancer like sequences in the cell types studied. Using a more stringent definition, the study by Andersson et al. in 2014 (PMID 24670763; see Further Reading) identified 43,011 putative enhancers in a detailed analysis of 800 human cell types, and characterized their activity across cell and tissue types.
DNA looping brings enhancers into close proximity to the promoters they control, probably with the assistance of the Mediator and cohesin complexes (Figure 5). Mediator, a 26-subunit protein complex, is involved in all aspects of gene expression; with its many subunits it can bind RNA polymerase and bridge across to transcription factors. Cohesin forms a ring that can enclose two DNA double helices. It is best known for its role in holding sister chromatids together but it is also involved in stabilizing promoter–enhancer interactions. Chromosome conformation capture confirms the reality of this looping, though it also reveals promoter–promoter and enhancer–enhancer interactions. As mentioned above, promoters and enhancers have a great deal in common. Unexpectedly, RNA sequencing reveals that enhancer sequences are transcribed, in both directions, producing short lived noncoding RNAs. The function, if any, of these eRNAs is unknown, but the study by Andersson and colleagues (2014) showed that they mark active enhancers. Possibly they help engage Mediator and assist enhancer–promoter looping. An important paper by Lupiáñez and colleagues in 2015 (PMID 25959774; see Further Reading) showed that enhancer–promoter interactions are controlled by the topologically associated domain (TAD) structure of chromosomes; changes in TAD boundaries cause mis-expression of genes.
Fig5. Enhancers and promoters are brought together by DNA looping. Regulatory elements such as enhancers (red box) may be located hundreds of kilobases upstream or downstream of the gene they control (blue box). DNA looping allows direct physical interactions between proteins bound to these distal elements and some of the many proteins bound to the promoter. The Mediator and cohesin multiprotein complexes stabilize the interaction. For clarity, only the RNA polymerase is shown at the promoter.
Enhancers are responsible for most of the tissue- and cell-specificity of gene expression. Important developmental genes are often located in gene deserts—long stretches of genomic DNA containing no protein-coding genes but many enhancers. Probably the presence of other expressed genes would excessively complicate the chromatin interactions necessary for enhancer function. A given gene may have a whole battery of enhancers (Figure 6). Individual enhancers control expression in specific tis sues, as revealed by experiments in which transgenic mice are made that carry a lacZ reporter gene driven by the enhancer in question. Some human diseases illustrate the role of tissue-specific enhancers. Total loss of function of an important developmental gene can cause multisystem abnormalities—but if just one enhancer is inactivated, while the other gene functions remain intact, the result may be just one particular facet of the usual syndrome (Table 1). The most important developmental genes, which specify the identity of cell lineages, are often associated with so-called super-enhancers or stretch enhancers. These are exceptionally long (>3 kb) and complex enhancer sequences that bind Mediator and a large number of transcription factors.
Fig6. Developmental genes may have several distant enhancers, each controlling one aspect of tissue specific expression. Arrows indicate the direction of transcription of the SHH (Sonic Hedgehog) developmental gene (first three exons shown) and of neighboring genes. Tissue-specific expression of the SHH gene is controlled by a series of enhancers (orange ovals) located up to 1 Mb upstream. The limb enhancer lies in an intron of the unrelated LMBR1 gene; point mutations in it produce polydactyly. Although these changes occur within the LMBR1 gene, that gene is actually irrelevant to the phenotype. tel, telomere. (Adapted from Lettice LA et al. [2011] Hum Mutat 32:1492–1499; PMID 21948517. With permission from John Wiley & Sons, Inc., © 2011 Wiley Periodicals, Inc.)
Table1. EFFECT OF INACTIVATING AN INDIVIDUAL ENHANCER OF A HUMAN DEVELOPMENTAL GENE WHOSE EXPRESSION IS CONTROLLED BY MULTIPLE ENHANCERS
Elongation of the transcript
Unlike DNA polymerases, RNA polymerases do not need a primer to get started, but further actions of the TFII transcription factors are needed before transcription can start. One subunit of TFIIH is a DNA helicase. This uses energy from the hydrolysis of ATP to open up the DNA double helix, giving RNA polymerase II (pol II) access to the template strand. TFIIH-dependent phosphorylation of serine residues in the C-terminal domain of pol II allows the polymerase to escape from the transcription start site and start RNA synthesis. However, two factors associated with the pol II protein, DSIF (DRB sensitivity-inducing factor) and NELF (negative elongation factor), cause it to pause 20–60 nucleotides down stream of the transcription start site. Short oligoribonucleotides are produced in a process of abortive initiation until further phosphorylation by P-TEFb (positive transcription elongation factor b) allows the polymerase to move into elongation mode. The average speed is then about 20 nucleotides per second, although there are pauses and spurts. At this speed it would take more than 24 hours to transcribe the 2.4 Mb dystrophin gene! Some data suggest that the transition from paused into elongation mode is the most critical part of transcription control. Most promoters, whether active or inactive, are said to be occupied by a pre-initiation complex, but only polymerases on active genes are able to move into elongation mode. As mentioned above, certain promoters carry both the activating H3K4me3 and repressive H3K27me3 marks. RNA polymerase at these promoters is said to be poised: it is thought that poised promoters may be particularly responsive to external signals.
Termination of transcription
Specific signals govern termination of transcription by polymerases I and III. For polymerase II, the only signal seems to be the poly(A) addition site—AAUAAA or a closely similar sequence. Two models have been proposed for how this causes termination.
• The allosteric model proposes that the polyadenylation signal induces a change in the RNA polymerase that commits it to termination.
• The torpedo model proposes that the polyadenylation signal triggers endonucleolytic cleavage of the nascent transcript some small distance downstream. The Xrn2 exonuclease attaches to the 5′ end of the RNA at the cleavage site, degrading it in a 5′ → 3′ direction. A sort of race ensues between the polymerase, continuing along the DNA and elongating the tail of the transcript, and the exonuclease coming up behind the polymerase and eating up the transcript (Figure 10.26). The exonuclease is the faster of the two, and when it catches up with the polymerase, transcription terminates.
Fig7. The torpedo model for termination of transcription. The transcript (red line) is cleaved (dotted line) downstream of the AAUAAA polyadenylation signal. An exonuclease attaches to the free 5′ end and works its way along the tail of the nascent transcript. Transcription ceases when the exonuclease reaches the body of the polymerase.
Porrua & Libri (2015) provide an extensive review, while Libri (2015) briefly summarizes recent evidence favoring each model (PMID 25650800 and 26474063, respectively; see Further Reading). The truth may lie in some combination of the two.
