NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 13 NUMBER 1 JANURARY 2006 5

Chromatin, transcript elongation and alternative splicingAlberto R Kornblihtt

A recent study reveals that the chromatin-remodeling factor SWI/SNF regulates alternative splicing by creating internal ‘roadblocks’ to transcriptional elongation where the phosphorylation status of RNA pol II is qualitatively changed.

The sequencing of the human genome revealed that it encodes a smaller than expected number of genes. This surprising observation has renewed interest in alternative precursor messenger RNA (pre-mRNA) splicing, as it is a mechanism to generate complexity of the proteome from a limited number of genes. In fact, alternative splicing affects the expression of 60% of human genes, and mutations that either create or abolish alternative splicing regulatory sequences, named splicing enhancers and silencers, are implicated in a wide variety of human diseases.

The regulation of alternative splicing depends not only on the interaction of splicing protein factors with splicing regulatory elements in the pre-mRNA, but also on the rate and pausing of transcriptional elongation. One of the most commonly observed alternative splicing events involves cassette exons that are not included constitutively in the mature mRNA. Such a cassette exon usually follows a suboptimal 3′ splice site that does not compete efficiently with a downstream exon that follows a strong 3′ splice site. To explain how RNA polymerase II(Pol II) elongation can affect the frequency of exon inclusion, I use a hypothetical gene that contains three exons, where the weak 3′ splice site precedes the second exon and the strong 3′ splice site precedes the third exon (Fig. 1). In the absence of internal stops to transcription or under high elongation rates, both 3′ splice sites are presented simultaneously to the splic-ing machinery. In this scenario (Fig. 1, left), the strong 3′ splice site could easily outcompete the weak one, resulting in exon skipping. However, if the polymerase slows down or pauses anywhere between these two 3′ splice sites, only the weak 3′ splice site is available for the spliceosome, and elimination of the first intron takes place. After the polymerase resumes transcription, the splicing machinery can then process the second intron, leading to the inclusion of the alternative exon (Fig. 1, right).

This kinetic coupling between alternative splicing and transcription elongation, usually referred to as a ‘first come, first serve’ mecha-nism, was originally suggested from experiments in which Pol II pause sites were artificially intro-duced into a gene, delaying the transcription of a splicing inhibitory element and therefore resulting in higher inclusion levels of one of its alternative exons1. A more direct proof for this mechanism was provided by the use of a mutant form of Pol II with a lower elongation rate. When transcription is carried out by this slow poly-merase, inclusion of alternative cassette exons in mature mRNA is greatly stimulated compared to inclusion during transcription by the wild-type enzyme2,3. The next question is thus, what regulates elongation and/or internal pausing to regulate alternative splicing? The answer seems to be related to the unexpected finding that promoter structure4,5 and recruitment of tran-scription factors6,7 and coactivators8 can greatly affect alternative splicing.

One of the ways in which promoter occu-pancy may affect alternative splicing is by elicit-ing specific Pol II phosphorylations that confer different elongation properties on the enzyme.

Indeed, the C-terminal domain (CTD) of the large subunit of Pol II contains a number of heptad repeats (52 in mammals and 26 in yeast) with a consensus sequence of YSPTSPS. The serine residues at positions 2 and 5 of these repeats (Ser2 and Ser5, respectively) are subject to regulatory phosphorylation. Phosphorylation of Ser5 by the basal transcription factor TFIIH is a mark for transcriptional initiation, whereas phosphorylation of Ser2 by the P-TEFb kinase complex promotes elongation. Accordingly, inhibitors of P-TEFb kinase such as DRB (dichlororibofuranosylbenzimidazole) pro-mote inclusion7, whereas transcription factors that stimulate elongation lead to skipping of alternative cassette exons6.

A second and less explored mode for the regulation of alternative splicing via elonga-tion involves chromatin structure. Changes in chromatin structure have been shown to affect splicing. For example, trichostatin A, a potent inhibitor of histone deacetylation, favors skip-ping of alternative exons7, presumably because hyperacetylation of core histones facilitates the passage of the transcribing polymerase. Furthermore, when transfected plasmids that

The author is in the Departamento de Fisiología y Biología Molecular, LFBM and IFIByNE-CONICET, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Ciudad Universitaria, Pabellón 2, (C1428EHA)Buenos Aires, Argentina.e-mail: ark@fbmc.fcen.uba.ar

Figure 1 The kinetic coupling model for the regulation of alternative splicing by Pol II elongation. The 3′ splice site (SS) preceding the alternative cassette exon (blue) is weaker than the 3′ SS of the downstream intron (red). Low transcriptional elongation rates (right) favor exon inclusion, whereas high elongation rates (left) favor skipping.

Promoter

CapCap

Cap (A)nSkipping

Constitutive exon Alternative exon Spliceosome

Inclusion

Pre-mRNAPre-mRNA

Pre-mRNA

mRNA Cap (A)nmRNA

Fast elongationNo internal stops

Slow elongationInternal stop

Pol II

Weak 3' SS Strong 3' SS

Pol II Pol II

N E W S A N D V I E W S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb

6 VOLUME 13 NUMBER 1 JANURARY 2006 NATURE STRUCTURAL & MOLECULAR BIOLOGY

are templates for alternatively spliced tran-scripts are prompted to replicate, a condition that favors chromatinization and consequently lowers elongation rates, skipping of alternativeexons is greatly reduced6. A study by Batsché et al.9 on page 22 of this issue reveals a new role of a chromatin-remodeling factor in alternative splicing. Surprisingly, the mechanism of action is independent of its chromatin-remodeling activity and involves the regulation of Pol II elongation.

SWI/SNF is a chromatin-remodeling factor that uses ATP to facilitate access to promot-ers by regulatory transcription factors. Found originally in yeast, its mammalian equivalent has been shown to interact with Pol II, splicing factors and spliceosome-associated proteins. Batsché et al.9 demonstrate that overexpression of Brahma (Brm), the ATPase subunit of SWI/SNF, favors inclusion of alternative exons of genes targeted by the SWI/SNF transcriptional activity, such as those encoding the adhesion proteins E-cadherin and CD44, the cell-cycle

regulator cyclin D1 and the pro-apoptotic fac-tor Bim. The catalytic activity of Brm does not seem to be necessary for this novel role in splic-ing; however, as expected for a splicing regula-tor, Brm is shown to interact with complexes containing the U1 and U5 small nuclear RNAs, which are present in spliceosomes, but not with the U3 small nuclear RNA, which is involved in ribosomal RNA processing. The authors find that Brm also interacts with Sam68, a nuclear RNA-binding protein previously shown to bind splicing regulatory elements present in the CD44 variable exons and to stimulate their inclusion upon activation of the ERK MAP kinases.

How does Brm use these multiple interac-tions to control alternative splicing? The authors use the CD44 transcript as a model to answer this question. The CD44 gene has a central block of ten consecutive variable (alternative) exons (v1–v10) between constitutive exons 5 and 16. Using chromatin immunoprecipitation (ChIP), Batsché et al.9 confirm that Brm not only is present at the promoter but also appears

distributed along the body of the gene with lev-els that decrease gradually toward the 3′ end. Although also concentrated at the promoter region, Pol II has a different distribution inside the body of the gene, with a clear accumulation within the variable region, peaking at exon v4. This peak disappears when endogenous Brm is knocked down by RNA-mediated interference but is higher when cells are treated with phorbolesthers, which activate ERKs. Consistent with this, Brm stimulation of inclusion of the CD44 variable exons is enhanced by the activation of ERK.

These findings suggest that activation of Sam68 by ERK triggers the formation of a macromolecular complex that contains Pol II and Brm. The nascent transcript may expose Sam68-binding sites at the central block of variable exons; the interaction between Sam68 in the macromolecular complex and the tran-script thus stalls Pol II at these sites. The stall-ing could favor inclusion of the variable exons in the mature mRNA, in agreement with the kinetic coupling model (Fig. 2). The study goes further to demonstrate that there is a dramatic change in the phosphorylation status of Pol II at the internal pausing site. Successive ChIPs (ChIP-reChIPs), using first antibodies to Brm and then antibodies to phospho-CTD for either phospho-Ser5 or phospho-Ser2, revealed that within the constant region Brm associates with phospho-Ser2 CTD Pol II. However, at the block of variable exons, Brm associates with phospho-Ser5 CTD Pol II species. Although entirely consistent with the kinetic model, this surprising observation breaks the dogma based on the assumption that Pol II containing phospho-Ser5 in its CTD is restricted to pausingsites immediately downstream of promoters, whereas elongating Pol II that is phosphorylated at Ser2 in its CTD distributes more or less uni-formly along the body of the gene. The appear-ance of a Pol II phosphorylation mark that is typical of initiation at specific sites within a gene could be either the cause or the consequence of the generation of internal ‘roadblocks’ to elonga-tion. It would be interesting to know whether reappearance of phopsho-Ser5 also accompa-nies other ways of generating internal pauses, such as the recently described intragenic DNA methylation that promotes the formation of a closed chromatin structure and subsequently reduces the efficiency of Pol II elongation at regions far downstream of the promoter10. In any case, it is now clear that internal roadblocks exist in vivo, can be regulated by external signals and are important for alternative splicing.

In view of the promoter effect and the role of Pol II elongation on alternative splicing, some years ago we proposed the idea that changes in the ‘pausing architecture’ of a gene would lead to

Figure 2 SWI/SNF stimulates inclusion of alternative exons in the CD44 gene by creating a ‘roadblock’ to Pol II elongation at the variable region. The pause is the consequence of multiple protein interactions involving SWI/SNF, Pol II, the splicing regulator Sam68 and spliceosomal components. The phosphorylation pattern of Pol II CTD is changed from phospho-Ser2 to phospho-Ser5, previously thought to be restricted to initiation at promoters. This might cause the stalling of Pol II molecules coming behind, even if they are phosphorylated at the elongation-competent Ser2.

P-Ser5 CTD Constitutive exon

Alternative exonSpliceosome Sam68

Pol II P-Ser2 CTDPol II

Pol

PromoterWithout SWI/SNF

With SWI/SNF

Cap

Cap (A)n

Pre-mRNA

Mature-mRNA

SWI/SNF

Pol II Pol II Pol IIPol II

Promoter

Cap

Cap (A)n

Pre-mRNA

Mature-mRNA

Pol II Pol II Pol II Pol II Pol II Pol IIII

N E W S A N D V I E W S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 13 NUMBER 1 JANURARY 2006 7

changes in the alternative splicing pattern of its transcript6. This idea seems to be confirmed by the study of Batsché et al 9. The powerful ChIP methodology has come of age as a tool to survey the ‘orography’ of the proteins involved in tran-scription, splicing11,12 and chromatin structure along the gene during cotranscriptional mRNA processing. One could imagine that, in a not-so-distant future, a detailed map of the peaks and valleys of the distributions of these proteins will be available for each gene in the genome. Such

a map would provide valuable information to predict gene expression patterns in vivo.

1. Roberts, G.C., Gooding, C., Mak, H.Y., Proudfoot, N.J. & Smith, C.W.J. Nucleic Acids Res. 26, 5568–5572 (1998).

2. de la Mata, M. et al. Mol. Cell 12, 525–532 (2003).3. Howe, K.J., Kane, C.M. & Ares, M. Jr. RNA 9, 993–1006

(2003).4. Cramer, P., Pesce, C.G., Baralle, F.E. & Kornblihtt,

A.R. Proc. Natl. Acad. Sci. USA 94, 11456–11460 (1997).

5. Kornblihtt, A.R. Curr. Opin. Cell Biol. 17, 262–268 (2005).

6. Kadener, S. et al. EMBO J. 20, 5759–5768 (2001).7. Nogués, G., Kadener, S., Cramer, P., Bentley, D. &

Kornblihtt, A.R. J. Biol. Chem. 277, 43110–43114 (2002).

8. Auboeuf, D. et al. Proc. Natl. Acad. Sci. USA 101, 2270–2274 (2004).

9. Batsché, E., Yaniv, M. & Müchardt, C. Nat. Struct. Mol. Biol. 13, 22–29 (2006).

10. Lorincz, M.C., Dickerson, D.R., Schmitt, M. & Groudine, M. Nat. Struct. Mol. Biol. 11, 1068–1075 (2004).

11. Gornemann, J., Kotovic, K.M., Hujer, K. & Neugebauer, K.M. Mol. Cell 19, 53–63 (2005).

12. Lacadie, S.A. & Rosbash, M. Mol. Cell 19, 65–75 (2005).

More than 1 + 2 in mRNA decappingSophie Bail & Megerditch Kiledjian

Decapping of messenger RNA was thought to involve a complex of only Dcp1 and Dcp2, but new data suggest that a larger multisubunit decapping complex exists in mammals. The larger complex includes a protein that facilitates the association of the two Dcp proteins and can be recruited by specific factors that promote mRNA decay.

Regulation of mRNA decay is an important step in the control of gene expression, and removal of the 5′ cap is a crucial part of eukaryotic mRNA decay. These mRNAs are not degraded by random processes, but rather undergo decay through two main exonucleo-lytic pathways that require the removal of the poly(A) tail (deadenylation) as a prerequisite step. The first pathway involves direct decap-ping and clearing of the mRNA body by the 5′➝3′ exonuclease Xrn1. The second path-way proceeds through a continuous degra-dation by the exosome complex followed by hydrolysis of the resulting cap carried out by the scavenger decapping enzyme DcpS1.

Two proteins, Dcp1p and Dcp2p, were initially isolated in yeast and shown to be essential for decapping, with Dcp2 being the catalytically active decapping enzyme2–4. In yeast, Dcp1p and Dcp2p are thought to form a decapping complex wherein the Dcp1p subunit facilitates decapping by Dcp2p5. Although homologs of Dcp1p are present in mammals, a similar stimulatory role has yet to be shown for either of the two human Dcp1p homologs, hDcp1a and hDcp1b. Furthermore, the composition of the decap-ping complex has also remained unclear.

A recent report by Fenger-Grøn et al.6 now suggests that hDcp1a and hDcp2 are components of a larger decapping complex. Using an affinity purification approach with hDcp1a as bait, three proteins were identi-fied. Two are homologous to previously identified yeast proteins known to enhance decapping, Dhh1p and Edc3p (termed rck/p54 and hEdc3, respectively, in humans). Consistent with previous results in yeast, both proteins were shown to specifically interact with hDcp1a. The third interacting factor corresponds to a protein of unknown function that was termed Hedls (pronounced “headless”). Hedls enhances mRNA decap-ping at least in part by facilitating the inter-action of hDcp1a with hDcp2. The authors propose that the presence of Hedls pro-motes formation of a multisubunit decap-ping complex that can consist of the two Dcp proteins, Hedls, rck/p54 and hEdc3 (Fig. 1). This is an important finding and suggests that Hedls can nucleate a larger decapping complex.

In contrast to yeast Dcp1p and Dcp2p, whichcan interact with one another5,7,8, minimal to no interaction was detected between over-expressed hDcp1a and hDcp2 unless Hedls wasalso expressed. Fenger-Grøn et al.6 also usedimmunopurified proteins to demonstrate a functional role for Hedls as a stimulator of hDcp2 decapping activity in vitro. Whether this is a direct effect of Hedls on hDcp2 or an indirect one via an interaction with

hDcp1a associated with rck/p54 and hEdc3 remains to be determined. The lack of Hedls in yeast indicates that yeast and mammals have diverged to use different mechanisms in decapping. Perhaps the more efficient association of Dcp1p with Dcp2p in yeast precludes the need for a Hedls-like pro-tein, whereas the more modest association of mammalian hDcp1a and hDcp2 requires

The authors are in the Department of Cell Biology and Neuroscience, Rutgers University, 604 Allison Road, Piscataway, New Jersey 08854-8082, USA.e-mail: kiledjian@biology.rutgers.edu

Decapping

(+) (+++)

Dcp2

Dcp1

Dcp2 Dcp1

Hedls

Rck

Edc3p54

Dcp2Dcp1

Figure 1 Model of potential Dcp2 decapping complexes. Top arrows represent the equilibrium of Dcp1 and Dcp2 association/dissociation in the presence or absence of Hedls. The decapping activity of Dcp2–Dcp1 complex (+) may be enhanced (+++) in the presence of the other decapping-complex components. The Dcp2 catalytic component is shown with teeth. Not shown are potential intermediates with only a subset of the proteins.

N E W S A N D V I E W S©

2006

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb