Minireview 855

A cap for all occasions Gabriele Varani

The 5¢ end of each polymerase II transcript is capped by amethylated guanosine triphosphate. The cap earmarks themRNA for subsequent processing and nucleocytoplasmictransport, protects the mRNA from degradation andpromotes efficient initiation of protein synthesis. Therecently solved structures of capping enzymes andcap–protein complexes shed light on how the 5¢ ends ofmRNAs are modified, and reveals the mechanisms bywhich the cap is recognized and how it functions in adiverse range of processes.

Address: MRC Laboratory of Molecular Biology, Hills Road,Cambridge CB2 2QH, UK.E-mail: gv1@mrc-lmb.cam.ac.uk

Structure 15 July 1997, 5:855–858http://biomednet.com/elecref/096921260050855

© Current Biology Ltd ISSN 0969-2126

Post-transcriptional regulation of gene expression in-volves the recognition by protein factors of specific sig-nals located throughout messenger RNAs. These signalsinclude distinctive terminal features, namely the N7-methylated guanosine triphosphate cap at the 5′ end andthe polyadenosine tail at the 3′ end. The cap, which hasthe same structure in all transcripts generated by poly-merase II, directs pre-mRNAs to the processing andtransport pathways in the cell nucleus and regulates bothmRNA turnover and the initiation of translation (Fig-ure 1). Understanding how the 5′ ends of mature mRNAs

are modified and recognized is therefore central tounderstanding how gene expression is regulated post-transcriptionally.

Capping enzymesThe modification of the mature 5′ end of an mRNAoccurs very early during transcription; in Drosophila, themajority of mRNA strands that have reached 30 nucleo-tides in length are already capped [1]. Capping occurs inthree steps: removal of the first phosphate by RNA tri-phosphatase to generate diphosphate-terminated pre-mRNA; capping with a GMP nucleoside mediated byRNA guanylyltransferase — a reaction which occursthrough an enzyme–GMP adduct intermediate; and N7-methylation by RNA (guanine-7)methyltransferase [2].Although all mono- or diphosphate terminated RNAs aregood substrates for guanylyl transfer, only polymerase IItranscripts are capped in vivo. This specificity is probablydue to interactions between the capping and the tran-scription machineries.

The crystal structure of the Chlorella virus guanylyltrans-ferase enzyme, PBCV-1, has remarkable similarities withDNA and RNA ligases. Sequence motifs that are con-served in cellular and viral capping enzymes and in DNAand RNA ligases cluster around the nucleotide-bindingsite in the PBCV-1 and T7 DNA ligase structures [3,4].The GTP–PBCV-1 and ATP–ligase interactions involvesimilar contacts between the nucleotide and residuesfrom these conserved motifs, whereas non-conserved

Figure 1

Nuclear membrane

(a) Co-transcriptional capping

(b) Pre-mRNA splicing

(c) Nucleocytoplasmic transport

(d) mRNA turnover

(e) Translational initiation

DNA

Pol II

Pre-mRNA

? Capping enzymes

?

CBC

Pre-mRNA

SnRNPs

mRNP

CBC Importin α

CBC

mRNA

Importin β

AAAAA

mRNA

Pab1p

AAAAAmRNA

Pab1p

?

Decapping enzymes 5′–3′ exonucleases Ribosome

eIF-4G eIF-4E

Multiple roles of the 5′ mRNA cap (red sphere) during gene expression in eukaryotes. CBC is the cap-binding complex made up of two subunits;Pab1p is the cytoplasmic poly(A)-binding protein; eIF-4E and eIF-4G are translation initiation factors. For details of each process (a–e) see text.

amino acids in the nucleotide-binding site define thespecificity of each enzyme [4]. Residues of the conservedmotifs line a groove between two structural domains(Figure 2). This cleft, which is probably the site of sub-strate binding, is much narrower in the capping enzymethan in the ligase; this probably reflects their differentspecificities for single-stranded RNA versus double-stranded DNA substrates.

The crystal structure of PBCV-1 guanylyltransferase re-vealed that it has two distinct conformations [4]. In thecrystal structure of guanylyltransferase, the deep cleft sepa-rating the two domains is ‘closed’ in one molecule in theunit cell by a rigid displacement of domain 2 (Figure 2).Upon soaking the crystals with divalent cations, a GMP–lysine adduct is formed, in which GMP binds within thesignature KXDG sequence [2] of the ‘closed’ but not of the‘open’ conformation. A strong electron density for GTP isobserved only for the ‘closed’ conformation, suggestingthat the catalytically active ‘closed’ structure is induced byGTP binding [4]. As the ‘closed’ form is not wide enoughto accommodate the RNA substrate, both conformationsare probably important for catalysis.

Cap-modifying enzymesSpecific classes of capped RNAs are subjected to furthermodifications. For example, the first nucleotide ofcertain viral mRNAs is methylated at the ribose 2′-OH,and the base N2 of the cap guanine is hypermethylatedin spliceosomal U snRNAs. A cap-specific 2′-O-methyl-transferase, VP39, from vaccinia virus recognizes thecanonical N7-methyl guanine cap structure and methy-lates the ribose hydroxyl group of the first transcribednucleotide. The VP39 structure is homologous to othermethyltransferases [5], and is superficially similar toRNA-binding proteins [6]. The fold of VP39 is distinctfrom typical αβ RNA-binding proteins, however, reveal-ing a novel architecture for RNA recognition. The βsheet of VP39 is not exposed to solvent as in RNA-binding proteins but buried in the protein core, whereasthe exposed surface is formed by loops and α helices(Figure 3) [5].

In the VP39–cap complex, the N7-methylated guanosineoccupies a deep aromatic cleft and is sandwiched be-tween the sidechains of Phe180 and Tyr22 (Figure 3) [7].These stacking interactions are likely to contribute to thespecific recognition of the methylated base through en-hanced π–π interactions, as observed for small aromaticmolecules. Thus, the unique electronic structure of themethylated base contributes to specific recognition of themRNA cap and (probably) of alkylated bases by DNAexcision repair enzymes.

856 Structure 1997, Vol 5 No 7

Figure 2

Superposition of ‘closed’ (yellow) and ‘open’ (green) conformations ofthe chlorella virus guanylyltransferase, with GTP (purple) bound to the‘closed’ state of the enzyme [4].

Figure 3

Structure of the vaccinia virus VP39 methyltransferase–cap complex,highlighting the stacking interactions between the modified guanine(purple) and essential aromatic amino acids (yellow) [7].

Pre-mRNA splicingThe role of the cap in mRNA splicing is mediated by anuclear cap-binding complex (CBC; Figure 1b), which iscomposed of two tightly associated proteins, CBP20 andCBP80 [8,9]. CBP20 is a member of the RNP superfamilyof RNA-binding proteins [6], but CBP80 has no clearhomology with other proteins. CBC (but not CBP80 orCBP20 in isolation) interacts specifically with the mono-methylated guanosine cap, both in vitro and in vivo [8,9],and promotes an efficient interaction between U1 snRNPand the 5′ splice site. This interaction between U1 snRNPand the 5′ splice site is one of the earliest steps in spliceo-some assembly [8] and is necessary to define the 5′ splicesite of the very first intron. This essential function of theCBC–cap complex and (probably) the mechanism by whichit promotes mRNA splicing are conserved between yeastand mammals, but the identity of the factor(s) thatmediate the interaction between CBC and the splicingmachinery and the structural basis for cap–CBC recogni-tion remain to be established.

Nucleocytoplasmic export and mRNA turnoverThe CBC associates with nascent transcripts and remainsbound to the cap throughout the splicing cycle, evenwhen the mature mRNA leaves the spliceosome [8]. Atthis stage, mRNA has to be packaged into a ribonucleo-protein particle (mRNP) for export to the cytoplasmthrough the nucleopore complex [10]. The cap–CBC com-plex is critical for the export of spliceosomal components(U snRNAs). The complex also plays a significant butnonessential role in mRNA export. Presumably, mRNPscontain multiple, redundant, export signals [9,10]. In yeastcells and Xenopus oocytes, the CBC forms an abundantcap-dependent complex with a component of the nuclearprotein import machinery (importin-α; Figure 1c), sug-gesting a link between nuclear RNA export and proteinimport [11]. Following mRNA transport to the cytoplasm,the CBC must dissociate from capped RNAs to allowbinding of a translation initiation factor, the cytoplasmiccap-binding complex eIF-4F. The CBC must then recycleback to the nucleus to function again in splicing and trans-port. Another factor involved in protein import, importin-β, causes the dissociation of the CBC–RNA complex(Figure 1c). The small GTPase RAN inhibits dissociationof the CBC–RNA complex in a GTP-dependent manner,suggesting a method of regulation which allows CBC to bereleased only after the RNA has been exported [11].

The turnover of mRNA is an important determinant ofthe levels and regulation of eukaryotic gene expression,and rates of decapping are primary determinants ofmRNA half-lives [12,13]. Any mRNAs that lack the capstructure or are prematurely decapped are rapidly de-graded in many eukaryotes by 5′–3′ exonucleases such asyeast XRN1 [13]. The 5′ cap and 3′ poly(A) tail functionsynergistically to regulate mRNA turnover; in the yeast

default pathway for mRNA degradation, decapping occursonly after de-adenylation. Yeast strains lacking Pab1p (thecytoplasmic poly(A)-binding protein) are decapped priorto de-adenylation [12], suggesting that Pab1p protects thecap from decapping enzymes (Figure 1d).

Translational initiationThe rate-limiting step in the initiation of protein synthesisrequires engagement between the 40S subunit of the ribo-some and the mRNA [14]. The cap plays a critical role inthis process, through the cytoplasmic cap-binding com-plex eIF-4F, again in synergy with the poly(A) tail (Figure1e) [15]. The cytoplasmic cap-binding protein, eIF-4E,specifically recognizes the cap and another component ofeIF-4F, the ‘adapter’ protein eIF-4G [14]. In yeast, eIF-4G also interacts with Pab1p [15,16], suggesting that stim-ulation of recruitment of the 40S subunit by the cap andthe poly(A) tail are mediated by the same mechanism, asillustrated in Figure 1e.

The requirement for the cap in translation is notabsolute. For example, internal ribosome entry sites ofsome cellular messenger and picornaviral RNAs bypassthe cap requirement. Those mRNAs that have a lowaffinity for eIF-4E, such as heat-shock gene transcripts,and mRNAs that are expressed during early develop-ment, are likely to rely on the poly(A) tail rather than thecap for translational initiation [16]. On the other hand,the translation of mRNAs in which very stable secondarystructures are present between the cap and the initiationcodon is strongly dependent on the eIF-4E–cap inter-action. The activity of eIF-4E is down-regulated byprotein inhibitors (4E-BPs) that block the eIF-4E–eIF-4G interaction. Translational control in response to growthfactors or mitogens is regulated through signalling path-ways, by changes in the relative affinities of eIF-4G and4E-BPs for eIF-4E [17,18].

Structures of the eIF-4E–cap complex reveal that eIF-4Ebelongs to the αβ family of RNA-binding proteins [6]; itforms a single αβ domain with a flat, exposed, antiparallelβ sheet, which is flanked by three α helices on the oppo-site surface [19,20]. The protein fold of eIF-4E is novel,however, being unrelated even to VP39 or either compo-nent of the nuclear CBC. Although the eIF-4E–cap inter-action is very different from those observed in GTPases(as was expected because the modified N7 cannot be usedas a hydrogen-bond acceptor), it is remarkably similar tothe VP39–cap complex [7]. When the cap is bound toeIF-4E, the alkylated guanine stacks between the side-chains of conserved tryptophans [19,20]. These π–π stack-ing interactions are of functional importance—mutationof each tryptophan to leucine abolishes binding, whereasmutations to phenylalanine reduces but does not abolishbinding. It has been known for some time that stackinginteractions are major determinants of RNA structure and

Minireview A cap for all occasions Varani 857

stability and of protein–RNA recognition [6]; it is nowclear that stacking also contributes to the recognition ofalkylated bases.

The futureThe diverse functions performed by the methylatedguanosine cap of polymerase II transcripts are trulyremarkable (Figure 1). The recent structures reviewedhere have provided needed insights into the mechanismof nucleotidyl transfer during capping and have revealedhow the cap is specifically recognized. To elucidatefurther the many functions of the 5′ cap, structures ofcomplexes of cap-binding proteins and associated initia-tion factors with RNA substrates are now needed.

Identifying the factors that couple capping, transcriptionand splicing is an important future goal, and understand-ing how the 3′ poly(A) tail and the 5′ cap functionallyinteract will help in explaining the details of mRNA sta-bility and translational initiation. It is also crucial tounravel the structural basis for the diverse protein–proteininteractions that mediate the role of the cap in so manyaspects of mRNA biogenesis, and to understand howthese interactions are regulated in response to develop-mental and environmental signals. Most encouragingly, asite on eIF-4E that binds a peptide motif, which is con-served between eIF-4G and 4E-BPs, has already beenidentified in the first of many structural studies [19].

AcknowledgementsIt is a real pleasure to thank Florante Quiocho (Baylor), Dale Wigley (OxfordUniversity) and their colleagues for sharing the coordinates of the structuresdetermined in their laboratories and Gerhard Wagner (Harvard) and col-leagues for sharing results before publication.

References1. Rasmussen, E.B. & Lis, J.T. (1993). In vivo transcriptional pausing and

cap formation on three Drosophila heat shock genes. Proc. Natl.Acad. Sci. USA 90, 7923–7927.

2. Shuman, S. (1995). Capping enzymes in eukaryotic mRNA synthesis.Prog. Nucl. Acid Res. Mol. Biol. 50, 101–129.

3. Subramanya, H.S., Doherty, A.J., Ashford, S.R. & Wigley, D.B. (1996).Crystal structure of an ATP-dependent DNA ligase frombacteriophage T7. Cell 85, 607–615.

4. Håkanson, K., Doherty, A.J., Shuman, S. & Wigley, B. (1997). X-raycrystallography reveals a large conformational change during guanyltransfer by mRNA capping enzymes. Cell 89, 545–553.

5. Hodel, A.E., Gershon, P.D., Shi, X. & Quiocho, F.A. (1996). The 1.85 Åstructure of vaccinia protein VP39: a bifunctional enzyme thatparticipates in the modification of both mRNA ends. Cell 85, 247–256.

6. Nagai, K. (1996). RNA-protein complexes. Curr. Opin. Struct. Biol. 6,53–61.

7. Hodel, A.E., Gershon, P.D., Shi, X., Wang, S.-M. & Quiocho, F.A.(1997). Specific protein recognition of an mRNA cap through itsalkylated base. Nat. Struct. Biol. 4, 350–354.

8. Lewis, J.D., Izaurralde, E., Jarmolowski, A., McGuigan, C. & Mattaj, I.W. (1996). A nuclear cap-binding complex facilitates association ofU1 snRNP with the cap-proximal 5′ splice site. Genes Dev. 10,1683–1698.

9. Izurralde, E., Lewis, J., Gamberi, C., Jarmolowski, A., McGuigan, C. &Mattaj, I.W. (1995). A cap-binding protein complex mediating UsnRNA export. Nature 376, 709–712.

10. Izaurralde, E. & Mattaj, I.W. (1995). RNA export. Cell 81, 153–159.11. Görlich, D., et al., & Izaurralde, E. (1996). Importin provides a link

between nuclear protein import and U snRNA export. Cell 87, 21–32.

12. Caponigro, G. & Parker, R. (1995). Multiple functions for the poly(A)-binding protein in mRNA decapping and deadenylation in yeast.Genes Dev. 9, 2421–2432.

13. Beelman, C.A. & Parker, R. (1995). Degradation of mRNA ineukaryotes. Cell 81, 179–183.

14. Hentze, M.W. (1997). eIF4G: a multipurpose ribosome adapter.Science 275, 500–501.

15. Tarun, S.Z., Jr. & Sachs, A.B. (1995). A common function for mRNA 5′and 3′ ends in translational initiation in yeast. Genes Dev. 9,2997–3007.

16. Tarun, S.Z.J. & Sachs, A.B. (1996). Association of the yeast poly(A)tail binding protein with translation initiation factor eIF-4G. EMBO J.15, 7168–7177.

17. Mader, S., Lee, H., Pause, A. & Sonenberg, N. (1995). The translationinitiation factor eIF-4E binds to a common motif shared by thetranslation factor eIF-4G and the translational repressors 4E-bindingproteins. Mol. Cell. Biol. 15, 4990–4997.

18. Brown, E.J. & Schreiber, S.L. (1996). A signaling pathway totranslational control. Cell 86, 517–520.

19. Matsuo, H., et al., Wagner, G., [Abstract]. Translational initiation factoreIF-4E: solution structure and interactions with mRNA cap and 4E-BP,Keystone Symposia — Frontiers of NMR in Molecular Biology, Taos,NM, February 1997.

20. Marcotrigiano, J., Gingras, A.-C., Sonenberg, N. & Burley, S.K. (1997).Cocrystal structure of the messenger RNA 5′ cap-binding protein (eIF-4E) bound to 7-Methyl-GDP. Cell 89, 951–961.

858 Structure 1997, Vol 5 No 7